Transcriptome of the inflorescence meristems of the biofuel plant Jatropha curcas treated with cytokinin

Effects of BA on Jatrophaflower development

To quantitatively apply BA to inflorescence meristems of Jatropha, absorbent cotton containing a BA solution was used rather than hand sprayers as in previous studies [10]. In line with our previous work [10], BA application to the inflorescence meristems of Jatropha resulted in a significant increase in the total flower number and the female flower number of each inflorescence (Figure 1A,B, Additional file 1: Figure S1A). In addition to promoting the number of normal female and male flowers, BA treatment also induced bisexual and asexual flowers (Additional file 1: Figure S1). As the number of female flowers is one of the most important factors affecting fruiting, the increased female flower number and the newly induced bisexual flowers contributed to more fruits than the control (Figure 1C,D, Additional file 1: Figure S1A).

454 sequencing and transcriptome assembly

For 454 transcriptome sequencing of Jatropha inflorescence meristems, four cDNA libraries were constructed from inflorescence meristems collected before BA treatment (Control), BA treatment for 2 hours (T), 4 hours post BA treatment (T +4 H) and 22 hours post BA treatment (T +22 H), respectively. Each cDNA library was used in a one-quarter run on a 454 GS FLX Titanium instrument, which produced a total of 839,205 raw reads (Table 1) that were handled by a 454 GS FLX system to cut off the adapters and low quality bases. Then, all raw reads were subjected to processing using an in-house-developed program to remove the low-quality reads. This procedure yielded 703,755 high-quality reads ranging from 100 bp to 790 bp, with an average length of 364 bp (Table 1). A total of 644,835 reads, corresponding to 91.63% of the high-quality reads, were assembled by the CAP3 program [47] using default parameters with an overlap of >40 bp and an identity of 90%. This analysis gave rise to 23,591 contigs with an average length of 740 bp, and 58,920 reads were identified as singletons (Table 1). Approximately 75.68% of the contigs were assembled from three or more reads. The distribution of the number of reads per contig, the length distribution of the contigs and the singletons are presented in Figure 2.

	0 HAT	2 HAT	6 HAT	24 HAT
Raw sequencing reads	200,913	201,598	248,674	188,020
Total number of raw read		839,205
Average length of raw read		351 bp
High quality reads	182,722	172,682	181,699	166,652
Total number of high quality read		703,755
Average length of high quality read		364 bp
Length range of high quality read		100 bp - 790 bp
Total number of singleton		58,920
Average length of singleton		334 bp
Total number of contig		23,591
Assembled reads (of total number of high quality read)		644,835 (91.63%)
Average length of contig		740 bp

Functional annotation

Functional annotation of CDSs was performed by searching against NCBI non-redundant protein database and KEGG protein database using BLASTP with E value 1e-3. Then the top hit protein (with the highest bit score) was chosen to count its origin (encoded by what organism). The distribution of sequences from various species closely related to those found in the Jatropha inflorescence meristem transcriptome is shown in Additional file 2: Table S1. More than half (58.4%) of Jatropha sequences are similar to sequences found in the close related castor bean (Ricinus communis) genome, which is consistent with the analysis of whole genome sequences of Jatropha[36, 37]. In contrast, the percentage of Jatropha sequences with top hit to sequences of the model plant Arabidopsis is as low as 0.81% (Additional file 2: Table S1). Given these results, studying the transcriptomes of Jatropha inflorescence meristems with and without BA treatment may shed light on the molecular mechanisms of cytokinin effects on the flower development of castor bean and other non-model plants.

Transcript clustering by expression signatures

To analyze the critical cellular processes in inflorescence meristems of Jatropha following BA treatment, we grouped the transcripts according to their expression patterns across four samples into eight clusters (Figure 3, Additional file 3: Table S2). Cluster I contains 517 transcripts down-regulated by BA treatment at T, T +4 H and T +22 H. Cluster II has 92 transcripts up-regulated by BA treatment at T, T +4 H and T +22 H. Cluster III is composed of 1,007 transcripts that are abundant at T, Cluster V is composed of 717 transcripts that are abundant at T +4 H and Cluster VII is composed of 1767 transcripts that are abundant at T +22 H. Conversely, Cluster IV is composed of 1,223 transcripts that are down-regulated at T, Cluster VI is composed of 2,629 transcripts that are down-regulated at T +4 H and Cluster VIII is composed of 1,370 transcripts that are down-regulated at T +22 H. A number of differentially expressed genes were included in more than one cluster according to their expression profiles. Thus although there are more than 9,000 genes presented in the Figure 3, there are only 5,871 genes in total expressed differentially following cytokinin treatment actually. These results revealed that there are more down-regulated genes than up-regulated genes after BA treatment. Rashotte et al. [48] found that more genes are down-regulated by BA than by zeatin treatment and inferred that some genes may be specifically down-regulated by BA that are not actually regulated by cytokinin. A recent transcriptome analysis of cytokinin responses in tomato leaves also showed that there were more repressed genes than induced genes early (2 h) in the response to BA treatment [34].

Compared to Arabidopsis orthologs, a number of Jatropha genes showed different expression patterns in inflorescence meristems exposed to cytokinin. The difference may result from different treatment methods and/or different concentrations of BA used for treatment. Most treatments of Arabidopsis are on seedlings rather than on inflorescence meristems employed in this study for Jatropha. And the concentration of BA usually was less than 20 μM in treatments of Arabidopsis[49], whereas the concentration of BA used in this study was 1 mM.

Gene ontology enrichment analysis of eight clusters

The enriched terms represented critical biological processes in inflorescence meristems treated with BA and indicated the cross-talk of cytokinin signaling with other biological processes, such as circadian rhythms [14, 60], glucose catabolism and nitrogen compound biosynthesis [52, 61].

Metabolic pathway analysis

Genes involved in cytokinin biosynthesis, metabolism and signaling

The differentially expressed genes that were annotated to function in cytokinin biosynthesis, metabolism and signaling were further analyzed.

Only one sequence in the transcriptome database, isopentenyl transferase 9 (IPT9), was annotated as functioning in cytokinin biosynthesis. Jatropha IPT9 (JcIPT9) was weakly expressed under both BA-treated and untreated conditions and was not responsive to BA treatment (Figure 4), which is consistent with previous results [52]. Four Jatropha LONELY GUYs (JcLOG3, 7, 8, and 9), which encode cytokinin-activating enzymes that function in the final step of bioactive cytokinin synthesis [64], were identified. JcLOG3, 7 and 8 also displayed weak expression across treated and untreated samples, whereas JcLOG9 was down-regulated in response to BA (Figure 4). Three Jatropha cytokinin oxidases/dehydrogenases (JcCKX1, 4, and 5) were present in the transcriptome database, and only JcCKX5 was up-regulated by BA treatment (Figure 4). The stable induction of CKX5 by cytokinin was confirmed in 9 of 13 microarray studies conducted by different labs [49].

Previous genetic and molecular studies suggested that cytokinin signal transduction occurs through a two-component system [65–69]. Transcripts involved in two-component systems were identified in Jatropha, and their expression patterns are shown in Figure 4. Three Jatropha orthologs of histidine kinases (JcHK2, 3 and 4), which act as cytokinin sensors [70], were identified. JcHK2 was slightly induced by BA treatment (Figure 4), the expression of which detected by qRT-PCR was significantly higher than control after BA treatment in this study (Figure 5). The other two receptors, JcHK3 and 4, showed no clear response to BA treatment revealed by both transcriptome and qRT-PCR analysis (Figures 4 and 5). The gene coding for the Jatropha ortholog of histidine phosphotransfer protein 1 (JcHP1) was significantly repressed upon BA treatment, whereas JcHP5 was not differentially regulated under our experimental conditions (Figures 4 and 5). Neither HP1 nor HP5 responded to cytokinin treatment in Arabidopsis[52, 66, 71]. However, repression of tomato HP1 expression after BA treatment was reported by Shi et al. [34], who performed a transcriptome analysis of cytokinin responses in tomato leaves. Four Jatropha orthologs of type-A response regulator (JcRRAs) and two type-B RR genes (JcRRBs) were identified in this study. JcRRA3 and JcRRA5 were up-regulated by BA treatment (Figures 4 and 5), and in previous studies, RRA3 and RRA5 were robustly induced by cytokinin in Arabidopsis[49, 52, 72]. However, the mRNA levels of JcRRA9 and JcRRA17 showed no significant differences between the BA-treated and control samples (Figure 4). In contrast to the 454 transcriptome data, the expression level of JcRRA9 determined by qRT-PCR showed a significant increase at T +22 H compared to control (Figure 5). Moreover, the transcription levels of JcRRB2 and JcRRB18, which are B-type RRs, were not significantly affected by cytokinin (Figures 4 and 5), which agreed with previous studies on the differential expression of B-type response regulators in response to cytokinins [66, 72].

Cytokinin-regulated genes involved in the metabolism and signaling of other phytohormones

Signaling pathways of phytohormones in plants are considered to be interconnected in a complex network [73]. Transcriptome studies indicated that cytokinin and auxin mutually regulate their signaling factors and/or their metabolism to control plant growth and development [48, 74]. In this study, the Jatropha ortholog of AUXIN-RESISTANT1 (JcAUX1), which augments auxin’s chemiosmotic influx into cells [75], showed response to cytokinin with a decrease in transcript abundance (Figure 6). Additionally, the Jatropha ortholog of the gene encoding the F-box protein TRANSPORT INHIBITOR RESPONSE1 (JcTIR1), an auxin receptor [76], was also down-regulated by BA treatment at T +22 H (Figure 5). Similarly, the Jatropha ortholog of INDOLEACETIC ACID-INDUCED PROTEIN 14 (JcIAA14), belonging to the AUX/IAA gene family, was also down-regulated (Figures 5 and 6). However, the Jatropha ortholog of another auxin-inducible gene that encodes small auxin up RNA (JcSAUR) [77, 78] was slightly but not significantly up-regulated by BA treatment (Figure 6). The auxin response factors (ARFs), which bind to conserved DNA sequences called auxin-response elements (AuxREs) in the promoter regions of primary auxin response genes, act as transcriptional activators or repressors depending on the nature of their middle domain [79–81]. Jatropha orthologs of ARF1 (JcARF1, transcriptional repressor) and ARF5 (JcARF5, transcriptional activator) [82] were identified, and the mRNA level of JcARF1 was increased whereas that of JcARF5 was decreased (Figure 6). Moreover, ARF5 was reported to repress ARR7/15, which negatively regulate cytokinin signaling in the shoot stem cell niche of Arabidopsis[82], indicating that cytokinin-regulated auxin signaling may be realized through the interaction between ARR7/15 and ARF5. The expression changes of genes involved in auxin signal transduction upon BA treatment in Jatropha were more similar to those of tomato [34] than to those of Arabidopsis[52]. Taken together, these results indicate that intricate agonistic and antagonistic interactions exist between cytokinin and auxin in the development of inflorescence meristems in Jatropha. They also indicate differences may exist in the gene responses to cytokinin between Jatropha and the model plant Arabidopsis.

Following BA treatment, Jatropha orthologs of genes encoding a GA receptor, GA INSENSITIVE DWARF1 (JcGID1), and a DELLA protein, RGA LIKE 2 (JcRGL2), were up-regulated (Figure 6). A previous study of cytokinin response genes in Arabidopsis by Brenner et al. [54] also reported the induction of DELLA protein-encoding genes by cytokinin and reported that cytokinin treatment caused reduced expression of GA20 oxidase. However, in our study, BA did not cause changes in the transcript abundance of a Jatropha ortholog of GA20 oxidase (Figure 6). Earlier studies showed feedback regulation between GA content and GID1 gene expression [83], and the amount of DELLA proteins was found to be inversely related to the amount of bioactive GA [84]. A Jatropha ortholog of another DELLA protein-encoding gene, REPRESSOR OF ga1-1 (JcRGA1), was slightly up-regulated (Figure 6). The induction of these GA signaling genes by cytokinin in Jatropha inflorescence meristems suggests that cytokinin reduces GA activity by up-regulating suppressors of GA signaling genes and supports previous findings that GA and cytokinin exert antagonistic effects on many aspects of plant development [52, 85].

Jatropha orthologs of two ABA biosynthesis genes, ABA DEFICIENT 1 (JcABA1) and JcABA2, were found to not be affected by BA treatment (Figure 6). However, we identified three Jatropha orthologs of ABA receptors, PYR1-LIKE (PYL) [86], JcPYL1, JcPYL4 and JcPYL8, and one of them (JcPYL8) was significantly repressed by BA, whereas mRNA levels of JcPYL1 and JcPYL4 showed no clear response to BA treatment (Figure 6). In addition, a Jatropha ortholog of protein phosphatase 2C (JcPP2C), whose activity was inhibited by PYLs in response to ABA in Arabidopsis[86, 87], responded to BA with an increase of transcript abundance (Figure 6). Conversely, the expression of a Jatropha ortholog of calcium-dependent protein kinase 4 (JcCDPK4), which is an important positive regulator in CDPK/calcium-mediated ABA signaling pathways at the whole-plant level in Arabidopsis[88], was down-regulated significantly by BA treatment (Figure 6). The BA responsiveness of Jatropha orthologs of genes involved in ABA signaling indicated an antagonistic interaction between cytokinin and ABA, which is in line with earlier studies revealing that cytokinins inhibit ABA production [89, 90].

Cytokinin also interacts with ethylene [91]. Decreases were identified in the transcript abundance of the Jatropha orthologs of ethylene receptor 1 (JcETR1) and the gene encoding mitogen-activated protein kinase 6 (JcMPK6) (Figure 6), both of which are negative regulators of ethylene signaling [92, 93]. The Jatropha orthologs of two additional genes, ethylene-responsive transcription factor 1 (JcERF1) and ethylene insensitive protein 3 (JcEIN3), were also repressed by BA treatment (Figures 5 and 6). In Arabidopsis, ethylene-insensitive protein 2 (EIN2) and EIN3-binding F-box 1 (EBF1) are induced by ethylene [94, 95]. The Jatropha orthologs of both of these genes were up-regulated by BA treatment in the present study (Figure 6), indicating that positive regulators of the ethylene signaling pathway were induced by BA treatment. The expression profiles of these genes suggested that cytokinin may act partially by influencing ethylene metabolism and signaling genes [96].

Four Jatropha orthologs of genes involved in brassinosteroid (BR) signaling were identified. The expression levels of two of them, BRASSINOSTEROID-INSENSITIVE 1 (JcBRI1) and BRI1-ASSOCIATED RECEPTOR KINASE 1 (JcBAK1), which work together as a receptor kinase pair initiating the signal transduction cascade [97, 98], were not significantly altered by BA treatment (Figure 6). The Jatropha ortholog of BRASSINOSTEROID-SIGNALING KINASE 8 (JcBSK8), which is in a small family of kinases that activate BR signaling downstream of BRI[99], was induced by BA (Figure 6). Furthermore, BA treatment inhibited the expression of the Jatropha ortholog of BRASSINAZOLE RESISTANT 1 (BZR1) (Figures 5 and 6), which is a transcriptional repressor with dual roles in BR homeostasis and growth responses [100]. These findings indicate positive crosstalk between the cytokinin and BR signaling pathways in Jatropha.

Genes encoding the jasmonate ZIM-domain (JAZ) proteins, which are key regulators of jasmonate signaling [101], have been reported to repress transcription of jasmonate-responsive genes [102]. The transcription level of the Jatropha ortholog of JAZ1 (JcJAZ1) increased upon BA treatment (Figures 5 and 6). The Jatropha ortholog of ENHANCED DISEASE SUSCEPTIBILITY 5 (JcEDS5) was up-regulated, whereas the Jatropha ortholog of NONEXPRESSOR OF PATHOGENESIS-RELATED GENES 4 (JcNPR4) showed no response to BA treatment (Figure 6). Both of EDS5 and NPR4 are involved in salicylate signaling [103]. These results imply that jasmonate and salicylate signaling may involve crosstalk with cytokinin in Jatropha.

Genes involved in flower development

BA treatment profoundly affected Jatropha flower development (Figure 1). The expression profiles of Jatropha orthologs of 23 flowering-related genes identified in Jatropha inflorescence meristems are presented in Figure 7A. And the relative expression of 14 of these genes were further confirmed by qRT-PCR (Figure 7B). The mRNA level of the Jatropha ortholog of CONSTANS-LIKE2 (JcCOL2), which contains a CCT (CONSTANS, CONSTANS-LIKE, and TIMING OF CAB1) domain and is related to flowering in response to the photoperiod [104], was significantly decreased by BA treatment (Figure 7A,B). In contrast, expression of tomato COL (Solyc07g006630) in leaves and Arabidopsis COL (At5g15850) in CKX1-overexpressing plants was induced upon treatment with cytokinin [34, 52]. The transcription level of the Jatropha ortholog of CYP89A5, a member of the P450 gene family, increased after BA treatment (Figure 7). Genes in the P450 gene family in Arabidopsis and tomato were also induced by cytokinin [34, 48, 52], suggesting that CYP89A5 may be involved in Jatropha inflorescence development in response to cytokinin. The transcript annotated to encode the Jatropha ortholog of GIGANTEA (JcGI), a circadian clock-controlled gene that regulates photoperiodic flowering in Arabidopsis[105, 106], was significantly induced at T +4 H (Figure 7A,B). Li et al. [13] reported that LEAFY (LFY), a floral meristem identity gene, was up-regulated by the accumulation of endogenous cytokinin in Arabidopsis. However, in our study, the Jatropha ortholog of LFY (JcLFY) showed no response to BA treatment (Figure 7A,B). The Jatropha ortholog of UNUSUAL FLORAL ORGANS (JcUFO), which encodes an F-box protein required for normal patterning and growth in the floral meristem in Arabidopsis[107], was down-regulated by BA (Figure 7B). AINTEGUMENTA (ANT) has been proposed to act downstream of auxin and involve in floral initiation, growth and patterning in Arabidopsis[108, 109]. After treatment with BA, Jatropha ortholog of ANT (JcANT) was profoundly down-regulated (Figure 7A,B). In Arabidopsis, ant flowers have fewer and smaller floral organs and possess ovules lacking integuments and a functional embryo sac [110, 111]. However, with decreased mRNA level of JcANT, flowers were not shown abnormal phenotypes in Jatropha after exposing to BA. Furthermore, Jatropha orthologs of other flowering-related genes such as CLAVATA1 (JcCLV1), KNOTTED-LIKE HOMEOBOX (JcKNOX), transcriptional corepressor LEUNIG (JcLUG), and serine/threonine protein kinase ABRUPTUS (JcABR) were also down-regulated (Figure 7A).

The ABCE model was formulated from the analysis of floral homeotic mutants with organ identity defects in two adjacent whorls of the flower, and similar classes of mutants were described in both Arabidopsis and Antirrhinum, suggesting that the regulation of organ identity was highly conserved in evolution [112–114]. In this study, Jatropha orthologs of A-function genes APETALA 1,2 (JcAP1,2), B-function gene PISTILLATA (JcPI), C-function gene AGAMOUS (JcAG) and E-function genes SEPALLATA1,2,3 (JcSEP1,2,3) were significantly repressed, with an exception of one B-function gene JcAP3 that was shown to be up-regulated by BA treatment (Figure 7A,B). Along with the altered expression of these ABCE model genes, bisexual and asexual flowers were induced, but no abnormal floral organ development was found in Jatropha treated with BA (Additional file 1: Figure S1, [10]). In Arabidopsis, exogenous BA application resulted in abnormal flowers that resemble the phenotypes of mutants, clv1, ap1, ap2, and ap3[22, 115]. However, the increased mRNA levels of AP1, PI and AG in transgenic Arabidopsis, which resulted from the expression of AtIPT4 under control of AP1, also caused abnormal flower and floral organ development [116]. The different phenotypic changes in Arabidopsis[22, 115] and Jatropha occurring upon BA treatment (Additional file 1: Figure S1, [10]) likely resulted from the differential responses of flowering-related genes as revealed in this study and previous studies on Arabidopsis[12]. It is noteworthy that the relative expression of several genes involved in flower development were found to be significantly reduced after BA treatment by qRT-PCR analysis (Figure 7B), whereas their expression did not show significant difference compared to the control in transcriptome sequencing analysis (Figure 7A). This discrepancy probably resulted from the low background expression of these genes and/or the insufficient transcriptome sequencing depth. Normalizing the cDNA library and enhancing the depth of transcriptome sequencing may help to solve this problem.

Cytokinin-regulated genes involved in cell division

It was reported decades ago that cytokinin promotes cell division [117]. More recent work found that an elevated level of endogenous cytokinin in Arabidopsis, achieved by overexpression of ATP/ADP isopentenyltransferase 4 (AtIPT4) [13] or loss-of-function of CKX, promoted cell division of the inflorescence meristem and the flower meristem [118–120]. In this study, expression of the S phase marker histone H4[64, 121] was found to be significantly higher in BA-treated (2 h) samples than in controls (Figure 8A). We identified four additional genes related to cell division in the database. Two Jatropha orthologs (Contig14149 and Contig2870) of cyclin D3 (JcCycD3;1 and JcCycD3;2), a D-type plant cyclin gene that plays an important role in the G1-to-S phase transition [121], were significantly up-regulated by BA treatment (Figure 8A,B). In Arabidopsis, cytokinin activates cell division through induction of CycD3[121–123]. In addition, the Jatropha ortholog of cyclin A (JcCycA3;2), an A-type plant cyclin gene that was reported to be associated with cell division [124, 125], was also induced following BA treatment (Figure 8A,B). Arabidopsis TSO1 (AtTSO1) encodes a protein with conserved CXC domains known to bind DNA that is homologous to animal proteins that function in chromatin complexes [126, 127]. AtTSO1 is involved in cell division during the development of the inflorescence meristem [128, 129]. Moreover, mutant alleles of AtTSO1 that cause defects in flower and ovule development were identified [130, 131]. Subsequently, Andersen et al. [127] reported that mutations in the AtTSO1 gene disrupt the development of both male and female reproductive tissues. JcTSO1, a Jatropha ortholog of AtTSO1, was highly expressed at T +22 H in Jatropha inflorescence meristems (Figure 8A,B). Based on this finding, we suggest that JcTSO1, together with other BA up-regulated and cell division associated genes (Figure 8), may be involved not only in promoting cell division in inflorescence meristems that resulted in increased total flower number per inflorescence after BA treatment (Figure 1, Additional file 1: Figure S1), but also in flower organ development that resulted in induced bisexual flowers upon BA treatment (Additional file 1: Figure S1, [10]).

Plant material and BA treatment

Cuttings from one Jatropha plant were propagated into individual plants and used as experimental plants. BA treatment was performed when the plants were undergoing the flowering stage in the following year. In total, 230 inflorescence meristems with a diameter of approximately 0.5 cm were selected as experiment subjects. First, each meristem was wrapped around by a piece of absorbent cotton weighing 10 mg. Then, 200 μl of a 1 mM BA solution containing 0.05% Tween-20 was applied to the cotton using a pipette.

Before BA treatment, 50 meristems were collected as control. Two hours after treatment, all cotton pieces wrapped around the inflorescence meristems were removed, and 50 meristems were sampled, which were identified as T (BA treatment for 2 hours). Fifty meristems were sampled at 4 and 22 hours post BA treatment, which were identified as T +4 H and T +22 H, respectively. The remaining 30 meristems were kept for phenotypic analysis. The fifty inflorescence meristems sampled at each time point, which were pooled together as one sample for 454 sequencing, were immediately frozen in liquid nitrogen and stored at −80°C.

To investigate the effect of cytokinin on branching of inflorescence, BA working solution (0.5 mM) with 0.05% (v/v) Tween-20 was sprayed onto each inflorescence meristem (about 0.5 cm in diameter) and the surrounding leaves using a hand sprayer. Control inflorescence meristems were sprayed with distilled water containing 0.05% (v/v) Tween-20. Spraying was conducted once. Thirty inflorescence meristems were used for each treatment.

RNA extraction, cDNA synthesis and 454 sequencing

Each frozen sample was ground in a mortar with liquid nitrogen, and total RNA was isolated using TRIzol reagent (Invitrogen Corp., Carlsbad, CA) following the standard protocol. An Agilent 2100 instrument was used to check the RNA quality (RIN >0.7), and a NanoDrop spectrophotometer (ND-2000C, Thermo Fisher Scientific, USA) was used to quantify RNA concentration. Messenger RNA was further purified using a MicroPoly(A) Purist Kit (Ambion) according to the protocol. Double-stranded cDNA was synthesized from mRNA according to Ng's full-length cDNA synthesis protocol [140] with some modifications [141] and then fragmented to 300–800 bp. The prepared cDNAs were transformed into single-stranded template DNA (sstDNA) libraries using the GS DNA Library Preparation kit (Roche Applied Science). sstDNA libraries were clonally amplified in a bead-immobilized form using the GS emPCR kit (Roche Applied Science) and sequenced on the 454 Genome Sequencer FLX instrument.

Sequence assembly and annotation

The 454 transcriptome sequencing reads were first handled by a 454 GS FLX system, which cut off the adapter and low-quality bases. The reads were then filtered by an in-house-developed program to remove low-quality reads. The qualified reads were then assembled by CAP3 using the default parameters [47]. Open reading frames were identified using an in-house-developed program based on ‘GetORF’ from EMBOSS [142], and the annotation was performed through BLAST searches against the Swiss-Prot and GenBank databases with an E-value cutoff of 1E-3. Gene ontology analysis was performed using GoPipe through BLASTP against the Swiss-Prot and TrEMBL databases using an E-value cutoff of 1E-3 [143]. The metabolic pathway was constructed based on the KEGG database by the BBH (bi-directional best hit) method [144].

Analysis of differentially expressed genes

To estimate gene expression, the read number for each gene was first transformed into RPKM (reads per kilobase per million reads) [145], and differentially expressed genes were identified by the DEGseq (identifying differentially expressed genes from gene expression data) package using the method MARS (MA-plot-based method with random sampling model) [146]. We use p-value <0.05 and the absolute value of log2Ratio >1 as the threshold to judge the significance of contig expression difference.

We perform BLASTX (NT query to AA database) in TAIR WU Blast against TAIR10 Arabidopsis proteins using NO filters to identify the most highly similar genes involved in plant hormone signaling, flower development and cell division (Additional file 5: Table S4). The members of the two-component signaling pathway in Jatropha was named following the nomenclature reported by Heyl et al. [147].

Where G is any gene from the transcriptome database and RPKM_control is the RPKM of control.

We identified enriched GO terms and pathways for the differentially expressed genes using DAVID (the Database for Annotation, Visualization and Integrated Discovery). DAVID is a web-based bioinformatics application that systematically identifies enrichment for biological annotations based on large gene lists derived from high-throughput genomic experiments [50, 51].

Quantitative real-time PCR (qRT-PCR) confirmation

The expression profiles of 15 genes involved in plant hormone signaling, 14 genes associated with flowering and 4 genes involved in cell division were investigated by qRT-PCR to confirm the transcriptome data. The RNA samples used for qRT-PCR were isolated from tissues collected from the experiment used for constructing the 454 libraries, and from inflorescence meristems collected in a replicate experiment. cDNA for each sample was synthesized from total RNA using the PrimeScript RT reagent Kit with gDNA Eraser (Perfect Real Time) (Takara, Japan) according to the instructions. The primers used for qRT-PCR are listed in Additional file 6: Table S5. The PCR products were sent to BGI (Shenzhen, China) for sequencing using specific primers for sequence confirmation. Relative gene expression levels were detected using the SYBR Premix Ex Taq II (Tli RNaseH Plus) (Takara, Japan) according to the manufacturer’s instructions on a LightCycler 480 II (Roche, USA) instrument. The relative expression of each gene was calculated using the 2^-ΔΔC_T method [149]. All quantitative PCR experiments were repeated with three biological and three technical replicates per biological replicate.

Accession numbers

All of the sequences of the unigenes greater than 200-bp in length obtained from the transcriptome sequencing of inflorescence meristems in Jatropha have been deposited in the Transcriptome Shotgun Assembly (TSA) database, http://www.ncbi.nlm.nih.gov/bioproject/265802 (Accession: PRJNA265802;ID: 265802). Unigenes less than 200-bp in length were listed in Additional file 7: Table S6.