Introduction

Transcription factors play a crucial role in numerous life activities and usually regulate gene expression by recognizing and binding to specific DNA elements in the promoter region of targeted genes. The Nuclear Factor-F (NF-Y) TF family, also called heme activator protein (HAP) or CCAAT binding factor (CBF), has been found in all the sequenced genomes of eukaryotes (Cao et al. 2011; Laloum et al. 2013; Cagliari et al. 2014). As a complex, the NF-Y protein consists of three distinct subunits, namely HapB (NF-YA), HapC (NF-YB) and HapE (NF-YC), and each subunit is encoded by one or two genes in yeast and mammals. To function for NF-F complexes, the translocation of NF-Y subunits into the nucleus is required. Due to lack of a nuclear localization signal, HapC usually interacts with HapE to form a tight heterodimer in the cytoplasm (Thön et al. 2010; Hackenberg et al. 2012a). HapB has a nuclear localization signal but needs the combined protein surface provided by dimerization of HapC and HapE as found in Aspergillus nidulans, rat and yeast (Sinha et al. 1995; McNabb and Pinto 2005; Thön et al. 2010). It has been found that there is a piggyback transport of complete heterotrimeric NF-Y complex in Aspergillus nidulans (Steidl et al. 2004). It was recently proposed that AtNF-YA and AtNF-YC were imported into the nucleus while AtNF-YBs could be translocated into the nucleus after interaction with NF-YC (Hackenberg et al. 2012a) in Arabidopsis. Many studies have demonstrated that the NF-Y complex plays a critical role in regulating cell proliferation and early embryo development through interactions among subunits in yeasts and mammals (De Silvio et al. 1999; Siefers et al. 2009).

In plants, the distinctive character of the NF-Y gene family showed that each subunit was usually encoded by multiple members (Petroni et al. 2012). For example, there are 30 members (including 10 NF-YAs, 10 NF-YBs and 10 NF-YCs) in Arabidopsis thaliana (Siefers et al. 2009; Petroni et al. 2012) and 59 members (including 12 NF-YAs, 32 NF-YBs and 15 NF-YCs) in Glycine max (Quach et al. 2015). The expansion of NF-Y gene number in plants showed genetic redundancy or functional divergence, which might contribute to the combinatorial and flexible TF network, and therefore, regulate plant growth and development and response to environmental changes (Calvenzani et al. 2012; Petroni et al. 2012; Ren et al. 2016). Many studies have found that NF-Y genes are responsible for regulating diverse physiological processes related to plant growth and development, such as embryogenesis (Lotan et al. 1998; Kwong et al. 2003; Mu et al. 2013), photomorphogenesis (Myers et al. 2016), photoperiod dependent flowering (Ben-Naim et al. 2006; Hackenberg et al. 2012b; Petroni et al. 2012), seed germination (Petroni et al. 2012; Huang et al. 2015; Liu et al. 2016), and root elongation (Ballif et al. 2011; Sorin et al. 2014). In particular, Arabidopsis Leafy Cotyledon 1 (AtLEC1, AtNF-YB9) gene, the first cloned and identified member in the NF-Y family, has been shown to play a role in regulating embryogenesis (West et al. 1994; Meinke et al. 1994; Lotan et al. 1998; Lee et al. 2003). Usually, AtLEC1 is specifically expressed in the developing embryo. Loss-of-function of AtLEC1 resulted in abnormal cotyledons and the end of embryo development (West et al. 1994) whereas over-expression of AtLEC1 led to the formation of embryo-like structures in the leaf tissues of Arabidopsis (Lotan et al. 1998). (AtLIL, AtNF-YB6), a homolog of LEC1, shared high similarity to LEC1 in sequences, but showed different expression patterns and could partly replace the function of LEC1 during embryo development (Kwong et al. 2003).

Previous studies have found evidence that NF-Y members such as LEC1 and L1L are involved in regulating fatty acid biosynthesis. Overexpression of LEC1 could result in the increased accumulation of fatty acid and triacylglycerol (TAG), and increased seed oil content in A. thaliana (Mu et al. 2008), Brassica napus (Tan et al. 2011) and maize (Shen et al. 2010). NF-Y members can also interact with other transcription factors to regulate seed oil content or fatty acid composition. For example, NF-YA3 interacts directly with WRI1-1 (AP2/ERF), together with NF-YC2 and ABI5 (bZIP) to regulate the transcription of oil biosynthetic genes in oil palm (Yeap et al. 2017). Arabidopsis LEC1 and NF-YC interacts with other transcription factors such as LEC2 to activate the transcription of the FAD3 gene that encodes an omega-3 fatty acid desaturase and catalyzes linolenic acid biosynthesis (Mendes et al. 2013). The mechanism of NF-Y in regulating the storage material accumulation in developing seeds needs to be further elucidated.

Recent studies have revealed that NF-Y members are involved in regulating physiological processes in response to various abiotic or biotic stresses (Nelson et al. 2007; Hackenberg et al. 2012b; Ni et al. 2013; Zhang et al. 2015a, 2016; Swain et al. 2016; Zhao et al. 2017). The over-expression of the NF-YA5 (AT1G54160) and NF-YB1 (AT2G38880) in Arabidopsis enhanced growth and survival under drought stress (Li et al. 2008). The heterologous expression of GmNF-YA3 from soybean exhibited a stronger drought tolerance in Arabidopsis (Ni et al. 2013). Similarly, the heterologous expression of Picea wilsonii PwNF-YA3 significantly enhanced the ability of Arabidopsis to tolerate salinity and drought stresses (Zhang et al. 2015b). These studies showed that the NF-Y family participated in various regulatory processes during plant growth and development with functional divergences among the members. However, only several genes were functionally documented in limited model plants such as Arabidopsis (Siefers et al. 2009), rice (Thirumurugan et al. 2008; Lee et al. 2015) and maize (Zhang et al. 2016).

The completion of genome sequences for several plants have allowed the comprehensive characterization and functional annotation of NF-Y genes in several species, such as Arabidopsis (Siefers et al. 2009), maize (Zhang et al. 2016), tomato (Li et al. 2016) and soybean (Quach et al. 2015). Castor bean (Ricinus communis L., Euphorbiaceae) is one of the most economically important non-edible oilseed crops and its seed oil is broadly used in industry due to its high ricinoleic acid content, which makes castor oil an ideal feedstock for biochemical and biodiesel production (Akpan et al. 2006; Ogunniyi 2006; Scholza and da Silva 2008). Castor bean is also extremely hardy: it is drought-tolerant, has a small gestation period and can adapt to a wide variety of soil conditions, including high levels of soil salinity. Due to increasing demand for production of castor bean seed oil in many countries, the development of improved varieties is of great importance to breeders and producers (Qiu et al. 2010). Further efforts should be made to elucidate the molecular mechanisms underlying the regulation of growth and development. The available genome sequences (Chan et al. 2010) provide a great opportunity for comprehensive characterization of the NF-Y family in castor bean, which could provide useful information in understanding the molecular mechanism of the NF-Y genes underlying the regulation of growth and development of castor bean. In the present study, genomic identification and structural characterization of the NF-Y family in castor bean was performed. The interactions among subunits comprising the NF-Y complex were assayed. Global expression profiles of the castor bean NF-Y genes in different tissues were examined using high-throughput transcriptomic sequencing technology and expressional responses to different abiotic stresses were inspected using the quantitative real-time PCR method. The results obtained here provide a comprehensive profile of the molecular basis of the NF-Y family in castor bean and other plants in the family Euphorbiaceae.

Materials and methods

Identification of RcNF-Y family members in castor bean

According to the updated classification criteria of NF-Y transcription factors in A. thaliana (Petroni et al. 2012), 30 amino acid sequences including 10 NF-YA, 10 NF-YB and 10 NF-YC subunits were retrieved from the TAIR (http://www.arabidopsis.org/). An initial search for NF-Y transcription factors in castor bean was performed using the protein sequences of Arabidopsis NF-Ys as queries against the protein database downloaded from the website http://castorbean.jcvi.org/index.php using the local software Blast + (2.2.24 +). Hits with significant E value (E ≤ 10−10) were extracted from the castor bean database using a custom perl script. Subsequently, the protein sequences were analyzed with the online SMART tool (http://smart.embl-heidelberg.de/) to confirm the presence of conserved domains in the NF-Y family. The resulting sequences were considered as NF-Y candidates in castor bean. The identified NF-Ys were named according to their order in the scaffold from top to bottom.

Alignments, phylogenies and gene structure analysis of RcNF-Ys

The multiple sequence alignment of each RcNF-Y family was carried out using full-length protein sequences in Clustal X software (Larkin et al. 2007). The protein alignments were imported into MEGA7 (Kumar et al. 2016) to construct phylogenetic trees using the neighbor-joining (NJ) method with 1000 bootstrap replicates. The gene structures of NF-Y members of castor bean and A. thaliana were created by comparing the full-length CDS (coding sequence) with the corresponding genomic DNA sequence, using the Gene Structure Display Server (GSDS, http://gsds.cbi.pku.edu.cn/) (Hu et al. 2015). In addition, the conserved motifs of all NF-Y members were investigated using the online MEME tool (http://meme.nbcr.net/meme/tools/meme). Parameters were set with optimum width of 10–200 amino acids, a motif with any number of repetitions, and the maximum number of motifs at 10.

Interactions between of RcNF-Y members by Y2H assay

To determine the protein–protein interaction between RcNF-YB and RcNF-YC or RcNF-YA members, we performed the yeast two hybrid (Y2H) assays. In this study, the constructs pDest22 and pDest32 (Invitrogen) were used for Y2H assays, which include transcription activating domain (AD) and DNA binding domain (BD), respectively. The CDS of each RcNF-YB gene was cloned into prey vector pDest22 through BP and LR recombination reactions, while the CDS of a single RcNF-YA or RcNF-YC gene was cloned into bait vector pDest32 in the same way. Two constructs of different combinations were co-transformed in yeast host strain Y2Hgold (Clontech) according to the previously reported protocol (Gietz and Schiest 2007). After co-transformation, yeast cultures were selected for leucine and tryptophan in synthetic drop-out medium (SD-Leu-Trp) (Clontech). Protein–protein interactions were screened by growth of at least three independent yeast colonies on SD medium without leucine, tryptophan and histidine (SD-Leu-Trp-His) and on SD medium without leucine, tryptophan, histidine and adenine (SD-Leu-Trp-His-Ade). All primers used in this experiment are listed in Table S1.

Gene expression analysis of RcNF-Ys in different tissues of castor bean

The RNA-seq data of five tissues types from the castor bean plant were downloaded from NCBI SRA under the accession ERA047687, including male developing flower, leaf, germinating seed and developing endosperms at stage II/III (free nuclear, early stage) and V/VI (later stage, cellular). The expression levels of RcNF-Ys were normalized to fragments per kilobase million (FPKM) and the heat map of gene expression was visualized by MeV 4.0 (http://www.tm4.org/mev.html). In addition, our previous transcriptomic data were also used to analyze RcNF-Y expression (Xu et al. 2013).

Stress responses of RcNF-Ys in castor bean

We treated two-week seedlings of castor bean ZB306 (provided kindly by Zibo Academy of Agricultural Sciences, Shandong, China) with four abiotic stresses: drought, cold, heat and salt for 12 h. The detailed procedures for each stress treatment were the same as described in the previous report (Xu et al. 2016). After treatment, the seedlings were immediately frozen in liquid nitrogen and used for RNA extraction. Total RNAs from different treatments were isolated using Tiangen RNAprep Pure Plant Kit (Beijing, China) according to the manufacturer’s instructions. The cDNAs were synthesized using PrimeScript™ II 1st Strand cDNA Synthesis Kit (Takara, Dalian, China) with DNase treatment. Quantitative real-time PCR (qRT-PCR) was performed with SYBR Green I fluorescent dye (Transgene, Beijing, China) in Bio-Rad CFX machine (Bio-Rad, Hercules, CA, USA). The PCR program was as followed: 95 °C for 2 min, 40 cycles of 95 °C for 30 s, 56 °C for 30 s and 72 °C for 30 s with a final dissociation stage. The internal reference gene, RcACTIN2, was used to normalize the expression levels of the tested genes. The primers used in qRT-PCR are listed in Table S2. Three independent biological replications were performed. Statistical analyses were carried out using the t test.

Results

The RcNF-Y family consists of 25 members

Through an extensive search, a total of 25 RcNF-Y members were identified from the castor bean genome, such as six RcNF-YAs, 12 RcNF-YBs and seven RcNF-YCs. They were designated as RcNF-YA1-6, RcNF-YB1-12 and RcNF-YC1-7 according to the recently suggested nomenclature rule for the NF-Y family (Siefers et al. 2009). All protein sequences contained the typically conserved domains of NF-Ys confirmed by SMART and Pfam online tools. Although additional six members had little similarities with NF-Y family, further analysis found that they should be classified as NC2s or Dbp3/4s, excluded in this analysis.

The basic information of RcNF-Ys and the corresponding homologs in Arabidopsis are shown in Table 1. RcNF-YA genomic sequences ranged in size from 2405 to 6870 bp and their coding sequence (CDS) lengths varied from 642 to 1053 bp, which resulted in amino acid numbers from 117 to 706. Their molecular weights (MW) were 23.44–38.14 kD and isoelectric points were 7.0–9.27. The genomic DNA sequences of RcNF-YBs varied greatly in size from 354 to 3808 bp when their CDS lengths were between 354 and 759 bp. RcNF-YB proteins had 117–252 AAs with MW of 13.34–27.97 and lower pI of 5.15–7.01. The sizes of RcNF-YC genomic DNA sequences were 354–6556 bp and the CDS lengths were 354–2121 bp. The RcNF-YC proteins had 117–706 AAs with pI of 5.26–8.37 and MW of 12.86–81.96. In addition, the locus of RcNF-YB11 contained two isoforms with variation in CDS length or gene structure. The primary form of RcNF-YB11.1 contained one intron within the gene regions and the CDS without the intron sequence encoded a protein with 117 AAs. The RcNF-YB11.2 isoform was intronless and the CDS which retained the intron sequence encoded a relatively large protein with 131 AAs.

Table 1 Basic information of 25 RcNF-Y members
Full size table

Phylogenetic analyses revealed the existence of three different clades in the RcNF-Ys family

To investigate the phylogenetic relationship of RcNF-Ys, the protein sequences of 25 members were aligned using Clustal X and phylogenetic trees were created by MEGA7 software with the neighbor-joining (NJ) criteria. Highly conserved domains were found among the members of each subunit as shown in Fig. 1, while N-terminal and C-terminal regions of RcNF-Y proteins showed more variable AAs. The core conserved domain of RcNF-YAs had about 53 AAs (Fig. 1a), which included two subdomains. Each subdomain contained 22 AAs, one for interaction with NF-YB/C and another for binding DNA. The central domain of RcNF-YBs had 95 AAs (Fig. 1b) and there were responding subdomains for DNA binding and for interactions of NF-YB/YC and NF-YB/YA. According to the distinct 16 AAs previously reported in the NF-YB conserved domain (Lee et al. 2003), RcNF-YB2 and RcNF-YB12 had obviously different residues from other 12 RcNF-YB members, and as a result two types of RcNF-YBs can be distinguished (LEC1 type and non-LEC1 type). The aspartate at D71 site was present in LEC1-type RcNF-YBs and was thought to be for protein interactions (Siefers et al. 2009). The central domain of RcNF-YCs contained about 80 AAs for interactions between subunits (Fig. 1c).

Fig. 1
figure 1

Multiple sequence alignments of castor bean NF-Y TF members. a Multiple alignment of RcNF-YA proteins. b Multiple alignment of RcNF-YB proteins. c Multiple alignments of RcNF-YC proteins. One member from Arabidopsis thaliana (AtNF-YA1, AtNF-YB1, AtNF-YC1) was included, respectively. Multiple alignments of NF-Y protein sequences were carried out with ClustalX and MEGA7. The identical AAs are marked with black shadows. The AA numbers of the conserved domains and the functional regions are marked

Full size image

The phylogenetic relationships of the members of each RcNF-Y subunit were analyzed and each subunit could be further clustered into two to three clades (Fig. 2). Among RcNF-YAs (Fig. 2a), four members formed a large group, whereas RcNF-YA1 and RcNF-YA6 were relatively distant to them. RcNF-YA2 and RcNF-YA3 showed the strongest similarities to RcNF-YA4 and RcNF-YA6, respectively. The members of RcNF-YBs can be divided into three groups (Fig. 2b) where the largest group consisted of seven members. RcNF-YB2 and RcNF-YB12 closely clustered together as LEC1-type. The RcNF-YCs formed three clusters (Fig. 2c). A further phylogenetic analysis was performed to explore the evolutionary relationships between NF-Y members from six species including moss (Physcomitrella patens), three dicotyledonous species (Phaseolus vulgaris, Arabidopsis, and castor bean) and two monocotyledonous species (Oryza sativa and Setaria italica). RcNF-Y members were clustered with their homologs from other species, suggesting that RcNF-Y members were conserved in evolutionary lines (Fig. S1). Notably, a distinct clade was identified as LEC1-type group in NF-YB, including RcNF-YB2, 12, AtNF-YB6, 9, PvNF-YB12, OsNF-YB7, 9 and SiNF-YB6, 9 (Fig. S1b). Based on previous studies (West et al. 1994; Lotan et al. 1998; Kwong et al. 2003; Mu et al. 2008; Tan et al. 2011), the members within this clade might be functionally conserved in plants, involved in regulating seed development and storage reservoir accumulation in developing seeds. It should be noted that the partial NF-YB and NF-YC members identified from moss (Zhang et al. 2015a) and Setaria italica (Feng et al. 2015), classified as NC2 and Dpb3/4, were included in this phylogenetic analysis (see Fig. S1b, c).

Fig. 2
figure 2

The phylogenetic trees of RcNF-YAs, RcNF-YBs and RcNF-YCs constructed by MEGA7 with NJ method and a bootstrap of 1000 replicates. a Phylogenetic tree of RcNF-YA members. b Phylogenetic tree of RcNF-YB members. c Phylogenetic tree of RcNF-YC members

Full size image

Characterization of gene structures and motifs

To identify potential functional regions in the NF-Ys protein, MEME tool was used to find out the conserved protein motifs of NF-Ys from castor bean and Arabidopsis. As a result, we found that the closely relevant NF-Y members between castor bean and Arabidopsis had highly conserved motifs and similar arrangements (Fig. 3). For 16 NF-YA proteins, all of them shared motif 4 and motif 5, and 12 members contained motif 9. In the case of 22 NF-YB members, three motifs (motif 1-3) were widely distributed. Four LEC1-type NF-YB members (including RcNF-YB2, 12 and AtNF-YB6, 9) shared the highly conserved motif 10. All of NF-YC members had the motif 1, and 15 NF-YC members contained motif 4, 6 and 7. We also observed that motif 8 was limited to a clade containing 10 members with abundant residues of glutamine (Q) in protein sequences, which was thought to play a critical role in transcription activation (Li et al. 2016).

Fig. 3
figure 3

Gene structures and conserved motifs of RcNF-Y and AtNF-Y proteins. Ten motifs were identified through MEME tool search and indicated with different colors. The blue box denotes untranslated region (UTR), yellow box represents exons within CDS while the line between them indicates intron. The length represents the corresponding size of exon and intron

Full size image

In additions, we compared the exon–intron structures of NF-Y genes of castor bean with Arabidopsis (Fig. 3). The gene structures of NF-YA members were highly conserved, consistent with previous studies of other species (Gusmaroli et al. 2001; Zhang et al. 2015a). For example, the NF-YA1 gene in castor bean contained three introns and its homologues in Arabidopsis had the same number of introns. The five RcNF-YA genes and their homologs in Arabidopsis shared four introns (except AtNF-YA5, which contained two introns). For NF-YB genes, 13 members were intronless, including five from Arabidopsis and eight from castor bean, whereas other members had variable numbers of introns from one to four. Among 17 NF-YC members, 12 members (including nine from Arabidopsis and three from castor bean) had no intron, and the number of introns within the other five members varied from one to five. The RcNF-YC2 member contained up to 15 introns. Although the intron length was variable between castor bean and Arabidopsis, five out of seven RcNF-Y genes and their homologs in Arabidopsis had the same intron numbers, suggesting that NF-Y genes were highly conserved in evolution (Liang et al. 2014).

Interactions between RcNF-Ys by yeast two hybrid assay

It was thought that NF-YB and NF-YC formed a tight dimer, which resulted in a new conformation that facilitated binding with NF-YA (Sinha et al. 1995; McNabb and Pinto 2005). Using Y2H assays, we detected the possible interactions between some RcNF-Y members. The constitutively expressed genes (such as RcNF-YB2 and RcNF-YC1) and most specifically expressed genes (such as RcNF-YB1, 7 and RcNF-YC4) were selected for further Y2H experiments. Each CDS of five RcNF-YAs and two RcNF-YCs were cloned into bait vector which contained DNA binding domain (BD) while each of six selected RcNF-YBs were cloned into prey vector which had activation domain (AD). As shown in Fig. 4, on SD-L-T-H, most of the combinations between RcNF-YB and RcNF-YC were interacted except RcNF-YA2 or RcNF-YA5, which were conjuncted with BD. On SD-L-T-H-A medium, five stronger interactions were observed in combinations between RcNF-YBs and RcNF-YCs and three interactions were found between RcNF-YAs and RcNF-YBs. No or less interactions were found in RcNF-YB3 or RcNF-YB4. Indeed, these results suggested that most of the RcNF-YBs and RcNF-YCs could interact each other.

Fig. 4
figure 4

Protein–protein interactions between RcNF-Y subunits by Y2H. RcNF-YB genes were fused to GAL4 Activation Domain (AD) and RcNF-YC or RcNF-YA genes were fused to GAL4 DNA-binding Domain (BD). SD-LT, SD-LTH and SD-LTHA stands for SD-Leu-Trp, SD-Leu-Trp-His and SD-Leu-Trp-His-Ade medium, respectively

Full size image

Overlapped and exclusive expression patterns of RcNF-Y members

To identify the potential functions of RcNF-Ys in the development of castor bean, we investigated the gene expression levels in five tissue types: developing endosperms at early (II/III) and later (V/VI) stages, germinating seed, leaf and male flower (Brown et al. 2012). The results showed that 22 of 25 RcNF-Ys were expressed in at least one of the five tissues, while three members including RcNF-YB5, RcNF-YB10, and RcNF-YB11 could not be detected (Fig. 5). There were 17 RcNF-Ys transcribed in leaf tissue, 14 in germinating seed, 16 in male flower, 18 in early endosperm (II/III) and 17 in later endosperm (V/VI) (Fig. S2). The transcription levels of each RcNF-Y gene varied across the different tissues. Among them, 13 genes comprising five RcNF-YAs, three RcNF-YBs and five RcNF-YCs were constitutively expressed in all tissues, nine genes were highly expressed in leaf tissue (RcNF-YA2, RcNF-YB8 and RcNF-YC5), and six genes (RcNF-YA5, 6, RcNF-YB9 and RcNF-YC4, 5, 6) were relatively abundant in the male flower. Notably, the gene RcNF-YB9 was abundantly and specifically expressed in the male flower. During endosperm development, as many as 17 RcNF-Y genes showed higher expression levels at the early stage than at the later stage. Six members (RcNF-YA2, RcNF-YA4, RcNF-YB2, RcNF-YB12, RcNF-YC4 and RcNF-YC6) were highly or specifically expressed in endosperms, and RcNF-YB2 and RcNF-YB12 belonged to LEC1-type members.

Fig. 5
figure 5

Expression patterns of RcNF-Ys in five tissues of castor bean. a Expression levels of RcNF-YAs. b Expression levels of RcNF-YBs. c Expression levels of RcNF-YCs. The heat maps were created according to transcriptomic data of five tissues, including leaf, male flowers, germinating seed and two endosperms at two different stages (II/III and V/VI) (Brown et al. 2012). Red represents higher expression levels while blue color indicates no transcript detected

Full size image

We further analyzed RcNF-Y expression patterns from the transcriptomic data of Xu et al. (2013), in which root tissues, seed1 (15 days after pollination) and seed2 (35 days after pollination) were included (Table S3). In total, 20 RcNF-Ys were found to be expressed in one or more tissues, including 16 in leaf, 14 in root, 14 in seed1, 15 in seed2 and 16 in endosperm (Fig. S2). There were 19 members that could be detected in both data sets, whileRcNF-YB10) and three genes (RcNF-YB8, RcNF-YB9, and RcNF-YC7) were missing in the data of Xu et al. (2013) and Brown et al. (2012), respectively (Fig. S2). Two genes (RcNF-YB5 and RcNF-YB11) could not be found in both sets of gene expression data. Most of the RcNF-Ys showed similar expression patterns where the same tissue was tested in both data. For most of the RcNF-Y genes, there were more transcripts in seed1 than seed2. Notably, RcNF-YB2 and RcNF-YB12 were only found in developing seeds and endosperm. Collectively, more than half of the RcNF-Y members were widely expressed in various tissues and several members were tissue-specifically expressed, such as seed-specific RcNF-YB2 and RcNF-YB12. As mentioned above, NF-Y members usually form a tight heterodimer or trimmer to activate the expression of targeted genes. To inspect whether the RcNF-Y genes interacted with each other, were validated by Y2H experiments and exhibited co-expression patterns among different tissues, we carried out analysis using the transcriptomic data of Xu et al. (2013). As shown in Fig. S3, the co-expressions of the RcNF-Y genes with protein–protein interactions were weak, suggesting the transcriptional regulations of different members might be independent from each other.

RcNF-Y members behaved differently in response to abiotic stresses

An increasing number of studies have demonstrated that members of NF-Y gene family are involved in plant responses to various abiotic stresses. In the current study, we examined changes in the gene expression levels of RcNF-Ys in two-week-old seedlings that were exposed to environmental stresses including drought, salt, cold and heat using the qRT-PCR method. The results showed that 19 RcNF-Ys were responsive to one or more abiotic stresses and RcNF-YC5 was little affected by any of the treatments tested (Fig. 6). Similarly, we found that five members (RcNF-YA1, RcNF-YB4, RcNF-YB5, RcNF-YB10 and RcNF-YC1) showed very low or no expression in the transcriptomic data, which could not be detected in two-week seedlings under the various abiotic stresses we tested.

Fig. 6
figure 6

Relative expression levels of RcNF-Ys in response to four abiotic stresses. The relative expression levels of five RcNF-YAs, nine RcNF-YBs and six RcNF-YCs were analyzed by qRT-PCR and RcActin2 was used as internal reference. The expression level of control without any treatment was set as 1.0. Each sample was performed with three repeats. The values represent mean value ± SD and the column with asterisk refers to statistical difference at P < 0.05

Full size image

After drought treatment, the expression levels of RcNF-YB12 decreased by up to 70%, while the expression levels of 13 RcNF-Y genes were induced. Four members (RcNF-YB1, RcNF-YB2, RcNF-YB8, and RcNF-YC6) were up-regulated and in particular, the expression level of RcNF-YC6 increased up to 16-folds (t test, P < 0.05). The expression levels of seven RcNF-Y members were up-regulated under the salt stress treatment, of which the largest changes were observed for RcNF-YB3, and RcNF-YC6. The expression levels of seven RcNF-Y members were down-regulated under the salt stress treatment. Cold stress increased the expression levels of six RcNF-Y members (especially RcNF-YC6), and decreased the expression levels of 9 genes (notably, RcNF-YC7). When castor bean seedlings were imposed to heat stress, the expression levels of 12 RcNF-Y members were elevated while those of five genes decreased. The expression level of RcNF-YB6 and RcNF-YC6 increased by up to a factor of 3, whereas the expression levels of RcNF-YB11 and RcNF-YC2 showed 90% decreases under heat stress (t test, P < 0.05). In addition, we noted that the expression levels of RcNF-YA2 and RcNF-YC6 were up-regulated while that of RcNF-YB12 gene was down-regulated by all four treatments. These results indicated that RcNF-Y genes were broadly involved in responses to various abiotic stresses.

Discussion

Although NF-Y transcription factors have been broadly studied in several plant species, the current study is the first report on genomic identification and characterization of the NF-Y transcription factors based on the genome of castor bean in the family Euphorbiaceae, an important group of resource plants. In total, according to the updated classification criteria of NF-Y transcription factors in A. thaliana (Petroni et al. 2012), 25 NF-Y genes were identified based on the castor bean genome sequences. Compared with the numbers of NF-Y members, such as 30 in Arabidopsis (Siefers et al. 2009), 33 in canola (Liang et al. 2014), 28 in rice (Thirumurugan et al. 2008; Petroni et al. 2012) and 30 in common bean (Rípodas et al. 2015), castor bean (genome size 310 Mb) harbored a comparable number of genes. However, soybean possessed as many as 66 (Quach et al. 2015) NF-Y members. These variations in gene numbers might be associated with variations in genome size. Also, these variations in gene numbers are related to the classification criteria, which include or exclude the partial NC2 and Dpb3/4 members based on the identification of Arabidopsis NF-Y (which included 36 or 30 members), e.g., maize and tomato had 50 (Zhang et al. 2016) and 59 (Li et al. 2016) NF-Y members, respectively, both of which include several NC2 and Dpb3/4 members.

The 25 RcNF-Y members identified in castor bean were divided into three subunits: RcNF-YA, RcNF-YB and RcNF-YC. Notably, RcNF-YB genes were separated into LEC1-type and non-LEC1-type, as found in both monocotyledons (Cao et al. 2011; Zhang et al. 2016) and dicotyledons (Siefers et al. 2009; Liang et al. 2014; Quach et al. 2015), suggesting the emergence of LEC1-type genes before the monocot–dicot split during plant evolution. Also, our results support the previous hypothesis that LEC1-type genes might originate from the non-LEC1 type (Cagliari et al. 2014). Each of the identified RcNF-Y subunits contained a conserved core domain which is responsible for DNA binding at CCAAT site (Fig. 1), suggesting that RcNF-Y transcription factors might share a similar regulatory pathway when they activate gene expression. Usually, the AA residues within NF-Y proteins are necessary to function. While comparing the AA residues for each subunit (RcNF-YA, RcNF-YB and RcNF-YC) based on amino acid sequences among castor bean, Arabidopsis (Siefers et al. 2009), canola (Liang et al. 2014) and maize (Zhang et al. 2016), we found that the AA residues were quite conserved in length (53 AAs for RcNF-YA, 100 AAs for RcNF-YB and 80 AAs for RcNF-YC). Previous studies have found that most of the members within NF-YAs possess 3–6 introns in genomes, whereas the intron/exon organization seems to be rather variable within NF-YBs and NF-YCs (Liang et al. 2014; Rípodas et al. 2015; Li et al. 2016; Ren et al. 2016; Zhang et al. 2016). Similarly, our current study found that most of the RcNF-YAs possessed four introns, while the number of introns between NF-YBs and NF-YCs was highly variable. In addition, based on analysis of gene structure and phylogenetic relationships, RcNF-YB2 and RcNF-YB12 were most likely to be the two LEC1-type orthologs of AtLEC1 and AtL1L that are pivotal in regulating the embryogenesis and seed development in Arabidopsis. Thus, we infer that RcNF-YB2 and RcNF-YB12 might share similar functions in regulating embryogenesis and seed development in castor bean, if these orthologs of LEC1-type genes are functionally conserved in plants.

A distinctive feature of NF-Y genes is that they usually function by formation of a heterodimeric or heterotrimeric complex. Our yeast two-hybrid experiments demonstrated that there were strong interactions between RcNF-YB and RcNF-YC members, whereas the interaction between RcNF-YA and RcNF-YB members was weak, consistent with the findings in Arabidopsis (Calvenzani et al. 2012; Hackenberg et al. 2012a). Since many heterogeneous combinations exist among NF-Y members in theory, only limited RcNF-Y complexes were verified in this study. In particular, we found that two NF-Y members (RcNF-YC4 and RcNF-YA2) were specifically or highly co-expressed in endosperm (Fig. 5), which exhibited strong interactions with another member RcNF-YB2 (specifically expressed in seed) in vitro, strongly implying that these factors most likely co-acted to participate in regulation of endosperm development and storage reservoir accumulation in castor bean seeds. Another constitutively expressed member RcNF-YB1 interacted with RcNF-YC1, which were co-expressed in diverse tissues including root, leaf, seed and male flower. In addition, previous research has shown that NF-Y protein can form diverse complexes with other proteins, such as basic region/leucine zipper motif (bZIP) bZIP67 (Mendes et al. 2013). The NF-YB/NF-YC dimers have been known to interact with a master regulator CONSTANS for regulating flowering (Ben-Naim et al. 2006). These findings indicate that fully investigating the heterogeneous protein interactions of the NF-Y family is a highly complex task.

NF-Y genes are functionally involved in regulating diverse physiological processes in plant growth and development (Zhao et al. 2017). Here, the expression level of more than half of the RcNF-Y members were identified from five tissues tested, suggesting that RcNF-Ys functionally participate in regulating the growth and development of different tissues. Various expression profiles of RcNF-Y members indicated their functional divergence. In this study, one of the main objectives was to identify the potential RcNF-Y members that regulate endosperm development and storage reservoir accumulation in developing castor bean seeds. In particular, we found that RcNF-YA2, RcNF-YA4, RcNF-YB12 and RcNF-YC6 were highly and specifically expressed in the early (endosperm II/III) and later (endosperm V/VI) stages of developing endosperm. This strongly suggests that these RcNF-Y members might participate in regulation of endosperm development and storage reservoir accumulation. Interestingly, RcNF-YB12, one of the AtLEC1 orthologs, as mentioned above, was highly and specifically expressed in the developing endosperm of castor bean. It has been well documented both that the gene AtLEC1 was specifically expressed in the developing embryo and that it acts to regulate embryogenesis (Lotan et al. 1998; Kwong et al. 2003; Mu et al. 2013). Our study adds to scientific understanding of the function of LEC1 in regulating not only embryogenesis but also endosperm development in the developing seed. Previous studies suggested that LEC1-type genes participated in regulating oil accumulation in developing seeds (Mu et al. 2008; Shen et al. 2010; Tan et al. 2011). In the present study, since the endosperm development and storage reservoir accumulation are an integrated process, it is difficult to separate the potential functions of RcNF-YA2, RcNF-YA4, RcNF-YB12 and RcNF-YC6 in regulating endosperm genesis or oil accumulation, although they were highly and specifically expressed in developing endosperm. Previous reports have shown that NF-Y members could interact with some transcriptional regulators such as bZIP67 or LEC2, which regulates fatty acid biosynthesis in Arabidopsis (Yamamoto et al. 2009; Mendes et al. 2013; Baud et al. 2016). It is, therefore, possible that these RcNF-Y members, which are characterized by high and specific expression in the developing endosperms, functionally participate in regulating oil accumulation in castor bean seeds. Further research is required to dissect the functions of these RcNF-Y members in regulating endosperm genesis or oil accumulation in castor bean. In addition, RcNF-YB9, an ortholog of AtNF-YB7 which usually functions in regulating flower development in Arabidopsis, was specifically expressed in male flowers. This is an indication of a promising direction for future studies of the function of RcNF-YB9 in regulating the flower development in castor bean.

As mentioned above, NF-Y genes were functionally involved in regulating plant responses to stress (Hackenberg et al. 2012a; Junker and Bäumlein 2012; Laloum et al. 2013; Petroni et al. 2012; Zhang et al. 2016; Swain et al. 2016; Zhao et al. 2017). Castor bean is highly tolerant of abiotic stresses (Qiu et al. 2010), but the potential molecular basis remains unknown. In this study, we showed that more than 50% of RcNF-Y members were responsive to abiotic stresses at the transcriptional levels, meaning that their functions might be involved in regulating plant responses to stresses. More studies are required to dissect the functions of RcNF-Y members in responses to stresses. In particular, expression levels of RcNF-YA2 and RcNF-YC6 were up-regulated, and that of RcNF-YB12 was down-regulated under all four treatments. When comparing changes in expression levels of NF-Y members in response to abiotic stresses between Arabidopsis (Hackenberg et al. 2012a) and castor bean, we noted that most of the NF-Y orthologs exhibited different or divergent expression profiles, though a few orthologs such as AtNF-YC2, RcNF-YB12 and AtNF-YB6 shared similar expression patterns. Not surprisingly, the functional divergence of most of the NF-Y orthologs in response to abiotic stresses varies widely across different species due to the variety of strategies and tolerance abilities of different plants to abiotic stresses. There is also evidence that NF-Y members function as a heterotrimeric complex in regulating plant response to abiotic stresses (Sato et al. 2014). It, therefore, seems important to identify the potential members that can form the heterotrimeric complex in regulating castor bean responses to abiotic stresses. The results of the current study provide a good opportunity for future studies to further identify potential heterotrimeric members according to the co-expression patterns with different abiotic stresses.

In conclusion, a comprehensive genomic characterization for the RcNF-Y family in castor bean was performed. The potential RcNF-Y members functionally involved in regulating the seed development and responses to abiotic stresses were characterized. The present study not only provides informative data to understand the regulation of seed development and storage reservoir accumulation in developing castor seeds, but also indicates potential RcNF-Y genes involved in regulating castor bean responses to abiotic stresses.

Author contribution statement

AL conceived and designed the study. YW, WX performed experiments and analyzed the data. ZC and BH participated in the data analysis. YW and AL wrote the manuscript. All authors read and approved the manuscript.