Identification and expression profiling of genes involved in circadian clock regulation in red dragon fruit (Hylocereus polyrhizus) by full-length transcriptome sequencing

ABSTRACT

Crassulacean acid metabolism (CAM) plants fix CO₂ at night, exhibiting a reversed regulatory pattern of metabolomic pathways compared with most model plants, which have C3 and C4 pathways. In this study, we used a valuable tropic fruit, red dragon fruit (Hylocereus polyrhizus), as model plant to identify and analyze the circadian regulation genes. Due to the absence of red dragon fruit’s whole-genome dataset, we established a full-length transcriptome dataset using single-molecule real-time (SMRT) sequencing method. A 7.66-Gb dataset with 4,552,474 subreads was generated, with an average length of 1,683 bp and an N50 of 2,446 bp. Using this dataset, we identified center oscillator genes: CCA1 (CIRCADIAN CLOCK ASSOCIATED1), ELF3 (EARLY FLOWERING 3), GI (GIGANTEA), LHY (LATE ELONGATED HYPOCOTYL), LNK1 (NIGHT LIGHT-INDUCIBLE AND CLOCK-REGULATED 1), and TOC1 (TIMING OF CAB EXPRESSION 1); a gene for the input pathway: CRY1 (CRYPTOCHROME); a gene for the output pathway: CO (CONSTANS); and genes related to the CAM pathway: MDH (MALATE DEHYDROGENASE), ME (MALIC ENZYMES), and PPDK (PYRUVATE PHOSPHATE DIKINASE). We further established the 24-h rhythmic expression pattern of these genes and classified these into three groups: HpCCA1, HpELF3, HpLHY, HpLNK1, and HpGI have expression peaks during the day; HpTOC1, HpCO, and HpCRY1 have highest expression levels at night; The genes involved in the CAM pathways, namely, HpMDH, HpME1, and HpPPDK, have double expression peaks in the day and night. Comparison of these expression patterns between red dragon fruit and model plants could provide clues in understanding the circadian clock regulation and the activity of the CAM pathways in cactus plants.

1. Introduction

Biological rhythm is a basic phenomenon in almost all life forms on Earth, occurring as periodic oscillations at different levels, from gene expression to metabolic regulation. Biological rhythm is regulated by circadian clocks, which are composed of three main components: the oscillator mechanism, input pathways, and output pathways.¹ These three parts of the biological clock act at different times throughout the day and night, and reciprocally regulate expression of circadian-related genes.²

The central oscillator, which is the core component of the circadian system, generates circadian rhythms. In Arabidopsis thaliana, the main feedback loop of the biological oscillator contains LHY, CCA1, and TOC1. The transcription factors CCA1 and LHY repress the transcription of the TOC1 gene in the morning via the feedback loop, whereas TOC1 represses the transcription of the CCA1 and LHY genes in the evening.^3,4 In addition, altering the expression of endogenous OsCCA1 under the control of the OsPRR1 promoter affects height and tiller number of rice (Oryza sativa).⁵ Feedback loops are a critical part of the oscillatory mechanism, their coupling with post-transcriptional, post-translational, and chromatin modifications also play an important role.^6,7

The input pathway transfers the environmental changes to the core oscillator of the biological clock by sensing light, temperature, and nutrition. Output pathways provide a bridge between the oscillators and the diverse biological processes.⁸ CRY1 is a blue light receptor that is involved in the regulation of photomorphogenesis and photoperiod regulation of flowering processes.⁹ Although GI is not considered a core clock component, it may connect a central oscillator to many physiological processes.¹⁰ The products of the GI and CO genes are important in the regulation of flowering in response to environmental stimuli.^10–12 The “evening complex” (EC), CCA1, and LHY regulate each other.¹³ EC indirectly promotes the expression of CCA1 and LHY, which, in turn, inhibit EC components.^14,15 LNK1 and LNK2 integrate early light signals with temporal information provided by core oscillator components.¹⁶ In addition, LNK1 and LNK2 play predominant roles in the regulation of flowering time and the inhibition of hypocotyl elongation compared to LNK3 and LNK4.¹⁷

Most of the current knowledge on circadian clock genes were collected from model plants such as Arabidopsis and rice. However, the metabolic features of CAM plants are the opposite of these C3 and C4 plants, i.e., CAM plants fix CO₂ in the dark using phosphoenolpyruvate (PEP) carboxylase and open their stomata at night for air exchange. CAM plants fix CO₂ to form oxaloacetate, which is then converted to malate by MDH enzymes. During the day, the cell stomata are closed, and after malic acid is released from the vacuole, pyruvate is produced under the action of ME, which is catalyzed by PPDK to form phosphoenolpyruvate.¹⁸ Previous research on CAM plants, including century (Agave americana), pineapple (Ananas comosus), white stonecrop (Sedum album), and milky widow’s thrill (Kalanchoë laxiflora), have revealed distinct regulatory mechanisms and different circadian rhythm expression patterns in CAM plants.^19–21 The rhythmic gene expression profile in the CAM plant Kalanchoë fedtschenkoi has been compared to that in Arabidopsis and revealed that the genes included in CAM pathways have phase-shifts between the two species, and only minimal overlap of rhythmic gene sets of each species has been observed.²² Red dragon fruit is a long-day tropical and subtropical fruit that is highly valued as a functional food, and supplementary lighting at night is widely used to induce flowering in short-day winter season and produce counter-season fruits to fulfill market demand.^23–25 Therefore, identification and analyzing the circadian genes in red dragon fruit not only will provide genetic evidence in understanding rhythm regulation in CAM plants but may also have potential agricultural applications.

However, because red dragon fruit cultivars have a very complicated heterozygous background.²⁶ No reference genome is currently available, which in turn severely impedes functional and molecular breeding studies. Third-generation transcriptome methods provide a possibility for mapping all expressed genes with full-length reading and relatively high accuracy. Recently, it has been used in understanding red dragon fruit development and synthesis of secondary metabolites.²⁷ In this study, we used the single-molecule long-read sequencing technology from Pacific Biosciences (PacBio) to obtain the full-length transcriptome dataset of red dragon fruit. Based on this dataset, we further detected and validated the rhythm of biological clock-related genes of red dragon fruit to provide genetic information for the circadian regulation of red dragon fruit.

2. Materials and methods

2.1. Plant material

Hylocereus polyrhizus cultivar Jindu 1was used in this study. To obtain a clear genetic background, 30-cm-long cuttings from a single two-year-old red dragon fruit were transported from the field to a climate room at 25°C with a 16 h:8 h (light:dark) rhythm, and 100 μmol photons m⁻²·s⁻¹ white-light illumination from an LED. The light was turned on at six in the morning. The stems and roots of this parent red dragon fruit were collected to extract RNAs for full-length sequencing. To synchronize their developmental stages, 5- to 6-cm-long new buds from transplanted cuttings were collected for rhythm gene analysis.

2.2. RNA extraction

Newly developed buds from red dragon fruit cuttings were collected every 2 h for 24 h. The collected tissues were frozen in liquid nitrogen and then stored at −80°C until use. Total RNA was extracted using a Plant RNA kit (OMEGA). RNA quality and quantity were determined by gel electrophoresis and a Thermo NanoDrop 2000, respectively.

2.3. PacBio Iso-Seq library preparation and sequencing.

Approximately 15 μg of mixed total RNA was reverse transcribed into cDNA, and the Iso-Seq library was prepared according to the isoform sequencing protocol (Iso-Seq) using the Clontech SMARTer PCR cDNA synthesis kit, and the BluePippin Size Selection System protocol as described by Pacific Biosciences (PN 100–092-800-03). The generated cDNA was reamplified by PCR. A Qubit fluorometer (Life Technologies, Carlsbad, CA, USA) was used to determine fragment size distribution. The quality of the libraries was assessed using the Agilent Bioanalyzer 2100 system. SMRT sequencing was performed using the Pacific Biosciences’ real-time sequencer using C2 sequencing reagents. The raw data were uploaded to Sequence Read Archive (SRA) (http://www.ncbi.nlm.-nih.gov/) as accession PRJNA659935.

2.4. Analysis of the full-length transcriptome

The analysis of the full-length transcriptome consisted of three stages: full-length sequence recognition, isoform-level clustering to obtain a consistent sequence, and a consistent sequence of polishing. First, the reads of insert (ROI) sequences extracted from the original depot sequence had the cDNA primers and polyAs in the sequence filtered, and then the sequence was divided into sequences based on the presence of the 3′ primer, 5′ primer, and polyA (optional), long and non-full-length sequences, and chimeric sequences and non-chimeric sequences. Then, the iterative isoform-clustering algorithm was used to cluster the full-length sequences from the same isoform, and the full-length sequences with similar sequences were clustered. A consistent sequence was contained in each cluster. Lastly, using the Quiver algorithm to cluster non-full-length sequences, the resulting consistent sequences were polished, and high-quality sequences were screened for subsequent analysis. Considering the limitations of a cDNA library, we screened the high-quality sequences because deletion of the 5ʹ end of a sequence in the library might indicate a non-full-length sequence. Therefore, we only pooled 5ʹ exon sequences, and the longest sequence was used as the final transcript sequence. To annotate the unigenes, we used the BLASTx program (http://www.ncbi.nlm.nih. gov/BLAST/) with an E value threshold of 1e⁻⁵ and searched the NCBI non-redundant protein (Nr) database (http://www.ncbi.nlm.nih.gov), the Swiss-Prot protein database (http://www.expasy.ch/sprot), the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (http://www.genome.jp/kegg), and the COG/KOG database (http://www.ncbi.nlm.nih.gov/COG).

2.5. Quantitative RT-PCR analysis

The full-length transcriptome sequence was obtained by SMRT from PacBio. The conserved sequence of the target gene based on the homology of the gene was amplified. Complementary DNA was synthesized using the FastQuant RT Kit with gDNase (TIANGEN). Quantitative RT-PCR analysis was performed using Roche LightCycler96 PCR and the ChamQ Universal SYBR qPCR Master Mix (Vazyme), and cDNA (equivalent to 10 ng of total RNA) was amplified using gene-specific primers in a 25-µL reaction volume according to the manufacturer’s instructions; each reaction had four replicates. QRT-PCR was performed twice using two independent RNA samples and generated similar results. All of the qRT-PCR primer pairs are listed in Table S1. To analyze the relative copy number of related genes, the housekeeping gene UBQ was used as positive nuclear control and reference for ΔCt calculations.

3. Results

3.1. Overview of the PacBio sequencing datasets

The transcriptome of 10 pooled samples was sequenced and analyzed with the PacBio Sequel platform to accurately capture full-length sequences. With SMRT, a total of 4,552,474 subreads (7.66 Gb) were obtained, with an average read length of 1,683 bp and N50 of 2,446 bp. To provide more accurate sequence information, a circular consensus sequence (CCS) was generated from reads that pass at least twice through the insert (Figure. S1a), and a total of 355,575 CCS were obtained (Table 1). By detecting the sequence, 260,718 were identified as full-length [containing the 5ʹ primer, 3ʹ primer, and the poly(A) tail] and 243,132 were identified as full-length non-chimeric (FLNC) reads with low artificial concatemers, and the mean length of FLNC was 2398 bp (Figure. S1b). The FLNC reads with similar sequences were clustered together using the ICE (iterative isoform-clustering) algorithm, and each cluster is considered as a consistent sequence by which we obtained 132,836 consensus isoforms (Figure. S1c).

Table 1.

Summary of reads from PacBio single-molecule long-read sequencing

	Subread	CCS	FLNC	Full length	5ʹ-primer	3ʹ-primer	Poly-A	Consensus reads
Number	4,552,474	355,575	243,132	260,718	311,755	318,519	301,999	132,836

3.2. Functional annotation of the red dragon fruit transcriptome

De novo assembly of these sequences was performed to obtain the unigenes. All corrected isoforms were functionally annotated by searching NR, Swiss-Prot, GO, NT, KOG, Pfam, and KEGG databases (Figure. S1d). GO analysis showed that the enrichment of 80,667 transcripts could be divided into three functional groups, namely, biological processes, molecular functions, and cellular components (Figure 1a). Homologous species were identified by comparing the transcript sequences to the NR database, and the results showed five species with the largest number of transcripts, namely, Beta vulgaris (33,884), Spinacia oleracea (18,002), Vitis vinifera (1,908), Hylocereus undatus (1,762), and Carnegiea gigantea (1,092) (Figure 1b). Genes in the biological process were mainly related to metabolic process, cellular process, single-organism process, biological regulation, localization, and regulation of biological processes. Genes involved in cellular component consisted of cell, cell part, macromolecular complex, membrane, membrane part, organelle, and organelle part. For the category “molecular function,” genes were mainly involved in binging, catalytic activity, and transporter activity. The KEGG results demonstrated that 83,638 transcripts were mapped to 365 KEGG pathways (Figure 2a).

Figure 1.

Functional annotation of corrected isoforms. (a) Distribution of GO terms for all annotated transcripts in the categories of biological process, cellular component, and molecular function. (b) Nr Homologous species distribution diagram of transcripts

Figure 2.

KEGG pathways enriched with transcripts

3.3. Identification and homology analysis of candidate genes

Based on transcriptional sequence annotation, we selected 11 clock-related genes as candidate genes to detect rhythmicity expression, i.e., HpCCA1, HpLHY, HpTOC1, HpGI, HpLNK1, HpCRY1, HpELF3, HpCO, HpME, HpMDH, and HpPPDK. Multiple isoforms were identified for each candidate gene (Table S2). These isoforms were then subjected to BLAST analysis using the online data bank Ensembl Plants (http://plants.ensembl.org/index.html), and selected isoforms with the highest alignment scores for each candidate gene were used for further analysis. We also analyzed the protein motif structures encoded by the selected genes from the red dragon fruit transcriptome dataset and their closest related genes in the MEME web server (http://meme-suite.org/index.html) (Figure 3). The protein motif structures in red dragon fruit were illustrated and compared to the closest protein structures from different species; for instance, HpCCA1 to Coffea canephora; HpLHY, HpTOC1, HpGI, HpLNK1, HpCRY1, and HpELF3 to Beta vulgaris; HpCO, HpME, and HpMDH to Actinidia chinensis; and HpPPDK to Citrus clementina.

Figure 3.

Motif composition of circadian clock-related genes compared to motif structure of known genes sequences with highest homologue scores. Black vertical lines indicate comparison pairs. The first line in each comparison pair is the protein name in red dragon fruit that was tested, and letters in brackets indicate the gene id encoded in the transcriptome database. The second line indicates the protein motif from the online database. Abbreviation of species: Ac: Actinidia chinensis; Bv: Beta vulgaris; Cc: Citrus clementina; Co: Coffea canephora; Hp: Hylocereus polyrhizus.

3.4. Rhythmic expression of genes related to circadian system

One isoform for each candidate gene was selected to examine the circadian expression pattern in red dragon fruit new buds, in which the developmental stages were synchronized (Figure 4a). We then compared sequences of each group of isoforms and selected the specific sequences as primers for qRT-PCR analysis to avoid potential interference from closely related isoforms (Table S1). At 2-h intervals, the 24-h rhythmic gene expression patterns of 11 genes were observed. These expression patterns can be divided into three groups: HpCCA1, HpELF3, HpLHY, HpLNK1, and HpGI have single expression peaks during the day (Figure 4b–f); the expression peaks of HpTOC1, HpCO, and HpCRY1 are observed at night (Figure 4g–i); and HpMDH, HpME1, and HpPPDK have two separate expression peaks, in the day and at night (Figure 4j–l).

Figure 4.

Rhythmic expression pattern of circadian clock-related genes in new buds of red dragon fruit. a) New buds with 5–6 cm lengths were used as samples; b)–l) Relative expression patterns in a 24-h period in the red dragon fruit buds. The abscissa is the sampling time, and 3 buds in each time point were collected as repeats (n = 3, error bar = standard error). This experiment was repeated twice and has identical results. The shadow represents night time, and the ordinate is the relative expression quantity

4. Discussion

Full-length transcriptome technique, also called third-generation RNA sequencing, is a useful and powerful tool in analyzing the gene expression pattern with obvious advances in sequencing of long RNA transcriptions.²⁸ A previous report has used the SMRT technique to analyze betalain biosynthesis pathways in red and white dragon fruit and also provided full-length RNA sequences in different development stages of fruits.²⁷ In addition to this database, we also report the full-length transcriptome database on vegetative organs of red dragon fruit cultivar Jindu 1. Table S2 shows that long-read isoforms for cDNA of genes of interest may also be detected in the transcriptome database (PRJNA659935).

We also determined that each identified gene has multiple isoforms in the full-length transcriptome dataset. Isoforms may reflect the diversity of the red dragon fruit genetic background because hybridization and bud mutations are widely used methods in red dragon fruit breeding.²⁶ Post-transcriptional editing and alternative splicing may also be other mechanisms that have generated isoforms.²⁹ However, the absence of whole-genome sequence data for red dragon fruit has prevented accurate analysis. To filter the isoforms of circadian regulation genes, we compared the sequence in the online data bank. The isoforms with the highest alignment scores in the data bank (http://plants.ensembl.org/index.html) were selected (Table S2). The conserved protein motif structures of the encoded proteins were compared with the most closely related proteins (Figure 3). The analogue of conserved motif structures confirmed that these selected genes can reflect their potential functions. Finally, the isoforms with high homology scores and analogue protein motifs to the known functional genes were selected for further analysis of rhythmic expression patterns.

At the core of the plant circadian oscillator, CCA1, LHY, and TOC1 comprised the first loop. At dawn, CCA1 and LHY are two Myb transcription factors acting at dawn that bind directly to the TOC1 promoter to negatively regulate its expression. In the evening, TOC1 acts a repressor of CCA1 and LHY,³ and the expression is promoted by RVE8 in the afternoon.³⁰ We observed that the expression of CCA1 and LHY begin to increase at late night, peaks at dawn, while the expression of TOC1 begins to decrease at night and then increase in the afternoon. These results imply that the red dragon fruit may have a similar feedback mechanism to the core oscillator in Arabidopsis. In red dragon fruit, we found that the expression peak of LNK1 occurs at noon, rather than shortly after sunrise, suggesting that LNK1 regulates the expression of afternoon genes in red dragon fruit.¹⁶ In addition, LNK1 is a coactivator of expression of clock genes TOC1 in Arabidopsis, and TOC1 repressed the LNK1 expression before dawn.^16,31 However, here we found a reversed expression pattern between HpTOC1 and HpLNK1, implying a different regulatory mechanism for morning gene expression in red dragon fruit.

Sensing of light rhythm and environmental temperature is crucial to the adjustment of flowering time of plants. CRYPTOCHROMEs are the blue light receptors that inhibit hypocotyl elongation and stimulate long-day floral initiation.³² To adjust the latter process, CRYs interact with COP1 and SPA1, causing the stabilization of CO and promotion of FLOWERING LOCUS T (FT).³² Red dragon fruit is a long-day plant that blooms at night, the expression of CO and CRY increases at night, and the expression peaks occur at night as well. These expression patterns may reflect the behavior of red dragon fruit flowering time compared to Arabidopsis and soybean that have expression peaks during the day.³³ CO stabilization is further regulated by GI in the afternoon by altering the ZTL-FKF1 complex.³⁴ This mechanism explains that when the expression abundance of GI gradually increases in the afternoon, the degradation of CO is inhibited. In addition, ELF3 and GI are essential because these enable the oscillator to synchronize the endogenous cellular mechanisms to external environmental signals. GI represses growth during midday to late afternoon, thereby contributing to the restriction of growth peaks in the morning.³⁵ ELF3 is an important component of the evening complex of the circadian clock and conveys environmental information to growth and developmental pathways as an input pathway gene.³⁶ The changing trend of ELF3 and GI is concordant with the finding of direct repression of GI by ELF3.^37,38

In pineapple, another typical CAM plant, the expression peaks of ME and PPDK occur in the light, suggesting that pyruvate decarboxylation and PEP phosphorylation occur during the daytime.¹⁸ Here, we observed double peaks in the expression pattern of HpMDH, HpME1, and HpPPDK, with higher peaks at daytime and lower peaks at night, concordant to the findings of previous studies that pyruvate decarboxylation, pyruvate phosphorylation, and malate dehydrogenation occur at light.³⁹ The MDH activity of white pitaya (Hylocereus undatus) grown in green house in Japan also showed two peaks at 4:00 and 12:00 in summer, while the PEPC and ME did not represent the similar rhythmic profile.⁴⁰ Even the regulation of the CAM photosynthesis pathways has been well studied, we did not find other reports describing double-peak expression patterns in MDH, ME1, and PPDK genes. Here, we focused on the rhythmic gene expression in the new buds of red pitaya, these findings provide new clues in understanding the regulation of CAM pathways in tissues with different developing stages. Based on the sequencing data provided in this manuscript, expression of related genes in different tissues can be further analyzed. Pitaya is one of the major economic crops in tropic regions and countries, illumination on the night is often used to justify its flowering and fruiting time. Analysis on the rhythmic expression genes will provide molecular evidences for the improvement of pitaya varieties.

Funding Statement

This work was supported by the Major Science and Technology Project of Hainan Province [311025]; Hainan Provincial Natural Science Foundation of China [2019RCI55].

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary materials

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