Detecting of chloroplast circular RNAs in Arabidopsis thaliana

ABSTRACT

Circular RNAs (circRNAs) are covalently closed single-strand RNA molecules identified in eukaryotes. Here we report 10 sequences of chloroplast genome (cpDNA) encoded circRNAs (cp-circRNA) from 3-day-old Arabidopsis thaliana seedlings via next-generation sequencing. Their parental sequences were referred to as exonic, intronic, and intergenic gene sequences. Due to the very low level of cp-circRNA expression, we attempted to use a digital polymerase chain reaction technique to validate their existence. Four of these circRNAs were confirmed: ath_circ_476, ath_circ_477, ath_circ_478, and ath_circ_480. Their expression levels in etiolated and de-etiolated seedlings were quantified using digital PCR. Of the four, the exonic circRNA ath_circ_476 was highly expressed in etiolated seedlings. Its parental gene, cytochrome C assembly protein (ycf5), has expression trends similar to ath_circ_476. Given that ycf5 is a single-exon gene, we propose that the circularization of ath_circ_476 occurs via an unknown process rather than via RNA-splicing-based mechanisms. Therefore, we provided a reliable method for detecting cp-circRNAs in low copies and implied an unknown circularisation mechanism of RNA in chloroplasts.

Introduction

Unlike linear RNAs, circRNAs are stable due to their closed-loop structures, with neither 5ʹ-3ʹ polarity nor poly-adenylated tails. With the development of high-throughput sequencing technology and computational analysis, circular RNAs have been found in eukaryotic and Archaea cells.^1-4 CircRNAs have crucial roles in the post-transcriptional regulation of gene expression in eukaryotic cells. For instance, circRNAs can act as sponges for microRNAs⁵ and circRNAs efficiently compete with miRNA for the recognition of RNA-induced silencing complexes (RISC), or circRNAs can regulate transcription and interfere with splicing.^6-8

CircRNAs in plant cells has also been sequenced and analyzed as well. CircRNAs are conserved, expressed at low concentrations, and are tissue-specific in A. thaliana.⁹ Sequence similarity analysis of the circRNAs of A. thaliana and functional annotation indicated that circRNAs are involved in many fundamental processes, including plant development, reproduction, and response to stimuli.¹⁰ CircRNAs derived from exon 6 of the SEP3 gene increase the abundance of cognate exon-skipping via stable R-loop formation, in turn driving floral homeotic phenotypes in Arabidopsis.¹¹ The genome-wide profiling of circRNAs revealed their widespread occurrence and potentially important biological roles in the transcriptional and post-transcriptional regulation in Oryza sativa.¹² Chen et al. suggested that the expression of circRNA and the parental gene are associated with the number of LINE1-like elements and their reverse complementary pairs (LLERCPs), and transposon-derived circRNAs likely modulate phenotypic variation via LLERCPs in Zea may.¹³ Using bioinformatics prediction and experimental validation, circRNA databanks have been established for O. sativa, A. thaliana, Z. mays, Solanum lycopersicum, and Hordeum vulgare. PlantcircBase is a database of plant circRNAs for the plant research community (http://ibi.zju.edu.cn/plantcircbase/). PlantcircBase contains databanks of circRNAs from all reported life forms of plants.¹⁴ The databanks include circRNAs encoded in cpDNA and mitochondrial DNA (mtDNA). However, the existence of cp-circRNA and mt-circRNA is supported only by bioinformatic prediction based on RNA-Seq results. We have little knowledge of the physiological functions of cp-circRNA and mt-circRNA, and their circularisation process remains unclear.

According to the widely accepted endosymbiotic theory, plastids and mitochondria originated from free-living prokaryotes, namely Cyanobacteria and Proteobacteria, respectively. Horizontal gene transfer among the genomes of plastids, mitochondria, and the nucleus occurred in the evolutional history of plants, resulting in the semi-autonomous genomes in plastids and mitochondria.¹⁵ Unlike the tightly packed mitochondrial genome (mtDNA), plastid genomes (cpDNA) have introns and a mechanism for the post-transcriptional regulation of gene expression.^16,17 The expression of cpDNA-encoded genes is often modified at the transcriptional and post-transcriptional levels, including cpDNA-specific cis- and trans-regulation elements, intron splicing, and RNA editing.^18,19 As circRNAs are non-coding RNAs originating from the intron-splicing process and act in the post-transcriptional regulation of gene expression, it is reasonable to investigate the cpDNA-encoding circRNAs in plants.

Northern blot and polymerase chain reaction (PCR) are mostly used to detect the circRNAs expression. Northern blotting has low sensitivity, but high specificity which is important for reducing false-positive results. Due to its time-consuming and laborious procedures, it remains an unpopular option. PCR is currently the fastest and easiest method to detect the expression of circular RNAs, such as qPCR and dPCR.²⁰ To determine the sequence and identify the backsplice junction of the circRNAs, the divergent primers were designed to flank the junction sequence and avoided amplifying random linear splicing events.^21,22 Digital PCR has shown great advantages and application prospects in the detection and identification of functional nucleic acids.²³ Different from qRT-PCR, dPCR is not depended on the cycle of threshold (Ct) for amplification curve, and is not easily affected by the amplification efficiency. Thus, dPCR is more reliable and accurate than qRT-PCR for determining the quantity of circRNAs with low expression level. Moreover, dPCR could achieve the absolute quantitative analysis.

In the process of skotomorphogenesis to photomorphogenesis, light has a crucial role for transfomation from etioplast to chloroplasts and photosynthesis initiation. Therefore, the expression of chloroplast circRNAs at the transcriptional level is important in the de-etiolated process. In this study, we performed circRNA assays on etiolated 3-day-old A. thaliana seedlings before/after de-etiolation by high-throughput sequencing and bioinformatics approaches. Sequencing libraries were generated using rRNA-depleted and RNase R-digested RNAs. Due to its low abundance so that we detected neither succeed with the RT-PCR nor the northern blot. Therefore, we attempted an accurate method to validate circRNAs in chloroplast by the digital PCR. Our results not only identified circRNAs in A. thaliana chloroplast but also provided a detection method for low-abundance circRNAs.

Material and methods

Plant materials

Seeds of wild-type (WT) A. thaliana from the Columbia 0 ecotype were planted on half-strength Murashige & Skoog medium (Duchefa Biochemie, the Netherlands) containing 0.4% Phytagel (Sigma, Germany) and kept in the dark at 4°C for 2 days. These vitalized seeds were illuminated under white light for 2 h and then incubated for 3 days at 22–24°C in darkness, namely etiolated seedlings. To convert into de-etiolated seedlings, the plants were exposed to white light (100 μmol/m²s) for 6 h, during which the yellow cotyledons were fully turned to green.²⁴ Whole seedlings were immediately frozen in liquid nitrogen and stored at −80°C until analysis.

RNA extraction

We used the E.Z.N.A total RNA kit (Omega, USA) to extract total RNA and remove the contaminating DNA from whole etiolated and de-etiolated seedlings. The quality and concentrations of the total RNA samples were assessed using 1.5% agarose gel electrophoresis and spectrophotometry.

Library preparation for circRNA sequencing

Five μg RNA per sample was used as input material. First, ribosomal RNAs were depleted using a Ribo-Zero™ rRNA Removal Kit (Epicentre, USA) to obtain rRNA-depleted RNAs. The rRNA-depleted RNAs were then treated with RNase R (Epicentre, USA) and subjected to TRIzol extraction. Sequencing libraries were generated using the rRNA-depleted and RNase R-digested RNAs with the NEBNext® Ultra™ Directional RNA Library Prep Kit for Illumina® (NEB, USA) following the manufacturer’s recommendations. Briefly, fragmentation was carried out using divalent cations at an elevated temperature in NEBNext First-Strand Synthesis Reaction Buffer. First-strand cDNA were synthesized using random hexamer primers and M-MuLV reverse transcriptase (RNase H). Second-strand cDNA was subsequently synthesized using DNA polymerase I and RNase H. In the reaction buffer, dNTPs with dTTP were replaced with dUTP. Finally, the library was purified (AMPure XP system) and then qualified using the Agilent Bioanalyzer 2,100 system (Figure S1).

Clustering and sequencing

The index-coded samples were clustered on a cluster generation system using a HiSeq PE Cluster Kit v4 (Illumina) according to the manufacturer’s instructions. After cluster generation, the library preparations were sequenced on an Illumina HiSeq 2500 platform, and 125-bp paired-end reads were generated. Sequencing was performed by Beijing Novogene Bioinformatics Technology (Beijing, China).

Mapping to reference genome

The reference genome and gene annotation were obtained from the genome website (A. thaliana TAIR10.30 ENSEMBLE). An index of the reference genome was built using Bowtie v2.0.6, and the clean paired-end reads were aligned to the reference genome using TopHat v2.0.9.²⁵

CircRNA identification

Unmapped reads were kept and 20-m from the 5ʹ and 3ʹ ends of these reads were extracted and aligned with the reference sequence independently using Bowtie v2.0.6.²⁵ Anchor sequences were extended using find_circ such that the complete read was aligned and the breakpoints were flanked by GU/AG splice sites. Then, the back-spliced reads with at least two supporting reads were annotated as circRNAs.⁶

Differential expression

Differential expression of samples without biological replicates was performed by using DEGseq (version 1.20.0).²⁶ P-value was adjusted by q-value,²⁷ and (q-value < .01) and (|log2(foldchange)| > 1) was set as the threshold for differential expression by default.

TaqMan probe and primer design

We obtained the sequences of the targeted circRNAs with RNA-Seq and the parental gene from the Arabidopsis Information Resource. The selected genes were used to design sequence-specific TaqMan probes and primers using Beacon Designer 8. The probes contained the back-splice site with the 5ʹ fluorophore FAM (6-carboxy-fluorescein) and 3ʹ fluorescent Quencher MGB (Minor Groove Binder) and had no consecutive single bases or palindromic sequences.²⁶ The divergent primers were located at both ends of the junction sequences and were designed in a ‘back-to-back’ manner with the 5ʹ ends of the forward and reverse primers.⁵ The lengths of the amplified fragments ranged from 50 to 100 bp. Control cDNA was used as a template to test pairs of primers by PCR to verify that they were usable. Table S4 lists the probe and primer pairs (synthesized in Thermo Fisher Scientific, USA) of the selected genes.

cDNA synthesis

cDNA was synthesized from 1 μg of total RNA using TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen, Beijing, China) in a 20-μL reaction mixture.

Validation with digital PCR

The first step in digital PCR is the dilution of the extracted nucleic acids to a concentration such that, on average, one template molecule is present per reaction well. PCR is then setup so that a multitude of such single-molecule PCRs is analyzed per sample. The Poisson distribution was used to estimate the average number of copies per reaction microliter. Thus, an individual reaction well could contain zero, one, or more template molecules. Digital PCR was performed on a QuantStudio™ 3D Digital PCR System platform consisting of a Gene Amp 9700 PCR machine (including a chip adapter kit), an automatic chip loader, and the QuantStudio™ 3D Instrument (Thermo Fisher Scientific, USA).

The total PCR volume was 15.5 μL and contained cDNA 5 μL, 7.5 μL of digital PCR Master Mix (Thermo Fisher Scientific, USA), 0.3 μL of probe, and 1.35 μL of the forward and reverse primers. The PCR program used a two-step process that was run at 96°C for 10 min; it was then run with 40 cycles of denaturation at 98°C for 30 s and annealing and extension at 60°C for 2 min; a final extension was run at 60°C for 2 min. Each reaction had three replicates.^28-30

The PCR was analyzed using QuantStudio 3D software (Thermo Fisher Scientific, USA). The statistical analysis was conducted using SPSS Statistics 18.0 software (IBM, Armonk, NY, USA). The significance of differences between dark-green and de-etiolated seedlings was analyzed using a t-test at the probability level of 0.05.

Results

Identification of circRNAs in A. thaliana

We analyzed circRNA assays on 3-day-old etiolated and 6 h de-etiolated A. thaliana seedlings using high-throughput sequencing technology and bioinformatic approaches. CircRNAs are recognizable based on backspliced reads in rRNA depleted and RNase R digested RNA-Seq data. The method used in animals⁶ to identify circRNAs in plants (Figure S1), resulting in a total of 8.50Gb of data (4.51 Gb from the etiolated sample and 3.99 Gb from de-etiolated sample), and our mapping exercise identified 57353650 unmapped reads in A. thaliana (Arabidopsis_thaliana TAIR10.30 ENSEMBL). After filtering out the contaminated and low-quality reads, 56693404 clean reads (98.00% of the total raw reads, Table S1). A total of 499 circRNAs and 4390 reads were identified in A. thaliana, using the strict threshold of at least two unique back-spliced reads count.

Sequencing of chloroplast circRNA in A. thaliana

The identified circRNAs were generated from all of the chromosomes, mtDNA, and cpDNA, respectively (Table S2). There were 10 circRNAs and 81 back-spliced reads on cpDNA, accounting for about 2% and 1.8% of the total numbers of circRNAs and backspliced reads. None of 10 cp-circRNAs was recorded in A. thaliana circRNAs from the plant circbase, whereas all their parental genes have been reported to have variate cp-circRNAs. Two (ath_circ_477 and ath_circ_481) had the canonical GT/AG splicing, whereas the other eight circRNA isoforms contained the splicing signals GG/AG, GA/AA, GT/GG, GC/TC, GG/TA, GC/AT, GG/GA, and AT/AG (Table 1). In total, 35 of 499 circRNAs splice sites were flanked by GT/AG formed by joining a splice donor to an upstream splice acceptor.

Table 1.

List of 10 chloroplasmid circRNAs identified by RNAseq.

CircRNA id	Strand	Parental genes	Spliced length	Type	Region	Backsplice site	dPCR verified
ath_circ_476	+	ATCG01040(YCF5)	159	Exonic	SSC	GG/AG	Yes
ath_circ_477	+		278	Intergenic	SSC	GT/AG	Yes
ath_circ_478	+		193	Intergenic	LSC	GA/AA	Yes
ath_circ_479	+		236	Intergenic	LSC	GT/GG	No
ath_circ_480	+	ATCG00400(TRNL.1)	512	Intronic	LSC	GC/TC	Yes
ath_circ_481	-		201	Intergenic	LSC	GT/AG	No
ath_circ_482	+	ATCG00500(ACCD); ATCG00490(RBCL); ATCG00540(PETA); ATCG00530(YCF10); ATCG00520(YCF4)	5115	Exonic	LSC	GG/TA	No
ath_circ_483	+	ATCG00640(RPL33)	201	Exonic	LSC	GC/AT	No
ath_circ_484	+	ATCG00720(PETB)	258	Intronic	LSC	GG/GA	No
ath_circ_485	-	ATCG00750(RPS11)	216	Exonic	LSC	AT/AG	No

The Circos plot revealed a non-random distribution of circRNAs in the chloroplast genome (Figure 1). The parental DNA sequences of cp-circRNAs were located mostly in the long single copy (LSC) region, and only two were located in the short single copy (SSC) region; there were none in the inverted repeat (IR) region (Figure 1). The 10 circRNAs could be categorized into three groups: 4 exonic, 2 intronic, and 4 intergenic circRNAs (Table 1, Figure 1). Among the exonic circRNAs, ath_circ_476, ath_circ_483, and ath_circ_485 were generated from the exons of single protein-coding genes, whereas ath_circ_482 was composed of exons from five genes. Among them, most of the exonic and intronic cp-circRNAs were encoded in the long single copy (LSC) region in the cp-DNA.

Figure 1.

The distribution of circRNAs identified in the A. thaliana chloroplast DNA using RNA-Seq. Red, green, and blue words represent exonic, intronic, and intergenic circRNAs, respectively. LSC: long single copy region; SSC: short single copy region; IR: inverted repeats regions.

Validation of the cp-circRNAs via digital PCR assay

Due to the very low expression rate of cpDNA-encoding circRNAs, none of their sequences has been confirmed by third-party methods. We attempted to use standard methods to verify the cp-circRNAs identified with the RNA-Seq technique, including Northern blotting and RT-PCR, but neither approach succeeded. Therefore, we further used digital polymerase chain reaction (dPCR) to confirm the expression of cp-circRNAs and quantify their absolute values. A set of sequence-specific TaqMan probes and divergent primers was designed and synthesized for cp-circRNA (ath_circ_476, ath_circ_477, ath_circ_478, and ath_circ_480). The probe sequence was complementary to the junction sequence of the target circRNA and contained the back-splice site (Figure 2a). Results indicated that, only a few fluorescence spots present on the chips for all 4 cp-circRNAs, whereas the chip with the blank control had no fluorescent spots (Figure 2b).

Figure 2.

Validation of cp-circRNAs in A. thaliana seedlings. (a) The design of the probes and primers. The probes contained the back-splice site with the 5ʹ fluorophore FAM and the 3ʹ fluorescent Quencher MGB, and the primers were located at both ends of the junction sequences. (b) The results of cp-circRNA in QuantStudio 3D. The blue points were positive and NTC was negative in etiolated and de-etiolated samples. Ath_circ_476, ath_circ_477, ath_circ_478, ath_circ_480, and ycf5 were identified.

Quantification of cp-circRNA in etiolated and de-etiolated seedlings

Both high-throughput sequencing and the dPCR assay can be used to identify the expression level of RNA molecules. Here, we compared the expression of the four confirmed cp-circRNAs between etiolated and de-etiolated seedlings using these two methods. The RNA-Seq technique gives the expression level as transcripts per million (TPM) (Figure 3a), and the dPCR technique gives the RNA concentration as copies per μL (Figure 3b). The RNA-Seq results showed that ath_circ_476 was expressed only in the etiolated seedlings, whereas ath_circ_477, ath_circ_478, and ath_circ_480 were expressed in the de-etiolated seedlings. The four cp-circRNAs were found in two seedlings using dPCR: ath_circ_476 was highly expressed in etiolated seedlings, not affected by light; the light regulation of ath_circ_478 and ath_circ_480 resulted in higher expression levels in de-etiolated seedlings; and ath_circ_477 was not differentially expressed between etiolated and de-etiolated seedlings. The dPCR expression patterns were consistent with the RNA-Seq results (Figure 3a,b), verifying the quality of our sequencing.

Figure 3.

Expression of cp-circRNAs and their parent genes in A. thaliana seedlings. (a) A histogram showing the RNA-Seq-quantified differential expression levels of four cp-circRNAs in 3-day-old etiolated seedlings and seedlings after 6 hof de-etiolation. TPM: transcripts per million. (b) A histogram showing the dPCR-quantified expression levels of four cp-circRNAs extracted from the same samples shown in A. (c) The different expression levels of the corresponding parent gene, ycf5, of ath_circ_476 in etiolated and de-etiolated seedlings by dPCR. Columns and error bars indicate the means and standard deviations of the relative expression levels (n = 3). *: Significant differences (p < .05).

Quantification of ath_circ_476 and its parental gene ycf5

To investigate the relationship between the expression levels of a circRNA and its parental genes, we analyzed the RNA abundance of ath_circ_476 and ycf5 using the dPCR technique (Figure 3c). The trace expression levels of ath_circ_476 were 0.83 and 0.22 copies/μL in the etiolated and de-etiolated seedlings, respectively, whereas there were 262,390 and 209,006 copies/μL, respectively, of the parental gene ycf5, or about 100,000–300,000 times more than for ath_circ_476. The differentially expressed parent genes shared a common expression trend with the corresponding circRNAs and were expressed more in etiolated than in de-etiolated seedlings (Figure 3c).

Discussion

According to the widely accepted endosymbiotic theory, plastids and mitochondria were originalized from the free-living prokaryotes, namely cyanobacteria and proteobacteria. Horizonal gene transfer (HGT) among genomes of plastids, mitochondria, and nucleus existed in the evolutional history of plants, resulting in the semi-autonomous genomes in plastids and mitochondria. Therefore, both semi-autonomous organelles maintained a miniaturized chromosome encoding protein which are involved in photosynthesis, biosynthesis, and genetic.^15-17 Numeric genes from their ancestor genomes were lost or transfered to the nuclear genomes. In other hands, some eukaryotic genes were transferred to the plastids genomes. Unlike the composite tightly packed mitochondrial genome, plastide genomes have introns, and the mechanism for the post-transcript regulation of gene expression.¹⁸ Since circRNAs are a non-coding RNAs origined from the intron-splicing process and act roles in the post-transcriptional regulation of gene expression, it is reasonable to investigate the cpDNA-encoding circRNAs in plants.

The development of chloroplasts and the expression of cpDNA encoding genes are a light-independent process. Nuclear genome encoding photoreceptors and light-depended biosynthesis of chlorophylls can initiate the signaling pathway to the de-etiolation process, i.e. the transform of etioplastids to chloroplasts. We used the next-generation sequencing technique and bioinformatic analysis to detect the cp-circRNAs from etiolated and de-etiolated arabidopsis seedlings. The loop portion of lariats may escape degradation by RNase R and even its reverse transcription product can also map to the reference genome. However, the junction reads of lariats are not contained GT-AG splicing signals, only a single dinucleotide GT in the 5ʹ end. Therefore, the circRNA predicted using the find_circ program such that the breakpoints were flanked by GU/AG splice sites in this paper. Ten different circRNAs were discovered in this study, non of them were recorded in current databases, while all of their parental genes have been identified in previous reports.¹⁴

Alternative splicing in RNA processing under different culture condition and development stages may cause this phenomenon.¹⁹ Moreover, cp-circRNAs have nine different back splicing pairs. Only two cp-circRNAs (ath_circ_477 and ath_circ_481) have the canonical splicing site identified among various eukaryotic with 5ʹ donor site of GT and 3ʹ acceptor site of AG. A total of 10 cp-circRNAs were identified and only two contained the canonical GT/AG splicing signals. These results demonstrated that base-pairs with non-GT/AG splicing signals are common and diverse for circRNAs in plants, coinciding to the early circRNA analysis based on rice.³¹ These results demonstrated that base-pairs of non-GT/AG splicing signals are common and diverse for cp-circRNAs. Furthermore, our results indicated the expression of intergenetic, intronic and exonic cp-circRNAs, coinciding with the published data of circRNAs encoded by nuclear genes. Among them, most of the exonic and intronic cp-circRNAs are encoded in the long single copy (LSC) region in the cp-DNA, according with the highly transcriptional activities in this region of cp-DNA.³²

The dPCR amplification partitioned an independent nano-PCR into the chip, the fluorescence each unit is individually measured and defined as positive or negative.^28,29 Among them, ath_circ_476, ath_circ_477, ath_circ_478, and ath_circ_480 were confirmed in the process from skotomorphogenesis to photomorphogenesis in seedlings. Ath_circ_476 and parent genes (ycf5) were significantly downregulated in and shared a common expression trend, suggesting co-regulation of circRNAs and their parent genes in etiolated seedlings. Interestingly, the parental gene of ath_circ_476, the ycf5, is a single exon gene. Therefore, the intron-driven circularization of RNAs is not suitable to explain the circularization of ath_circ_476. Our finding raised clues to investigate an individual mechanism of RNA circularization in chloroplasts.

Funding Statement

This work was supported by the National Natural Science Foundation of China [31671489].

Author contributions

SL did most experimental works and wrote the manuscript; GW did the dPCR analysis; YW supervised this project.

Conflicts of Interest

The authors declare no conflicts of interest.

Supplemental material

Supplemental data for this article can be accessed on the publisher’s website.