Genome-wide identification, phylogeny, and expression analysis of the SBP-box gene family in Euphorbiaceae

Background Euphorbiaceae is one of the largest families of flowering plants. Due to its exceptional growth form diversity and near-cosmopolitan distribution, it has attracted much interest since ancient times. SBP-box (SBP) genes encode plant-specific transcription factors that play critical roles in numerous biological processes, especially flower development. We performed genome-wide identification and characterization of SBP genes from four economically important Euphorbiaceae species. Results In total, 77 SBP genes were identified in four Euphorbiaceae genomes. The SBP proteins were divided into three length ranges and 10 groups. Group-6 was absent in Arabidopsis thaliana but conserved in Euphorbiaceae. Segmental duplication played the most important role in the expansion processes of Euphorbiaceae SBP genes, and all the duplicated genes were subjected to purify selection. In addition, about two-thirds of the Euphorbiaceae SBP genes are potential targets of miR156, and some miR-regulated SBP genes exhibited high intensity expression and differential expression in different tissues. The expression profiles related to different stress treatments demonstrated broad involvement of Euphorbiaceae SBP genes in response to various abiotic factors and hormonal treatments. Conclusions In this study, 77 SBP genes were identified in four Euphorbiaceae species, and their phylogenetic relationships, protein physicochemical characteristics, duplication, tissue and stress response expression, and potential roles in Euphorbiaceae development were studied. This study lays a foundation for further studies of Euphorbiaceae SBP genes, providing valuable information for future functional exploration of Euphorbiaceae SBP genes.


Background
Transcription factors (TFs) are DNA-binding proteins that play essential roles in the regulatory networks of critical developmental processes [1]. According to the specific protein structure, TFs can be divided into distinct families. SQUAMOSA promoter-binding protein (SBP)-box (briefly: SBP) or SBP-like (SPL) genes encode a type of TF family that is uniquely conserved in plants.
SBP genes were first identified in Antirrhinum majus, and they were found to regulate the expression of MADS-box genes, which are critical in floral development [2]. Since then, studies on SBP genes have continually been carried out. As a result, SBP genes have continually been identified in plants ranging from monocyte algae to flowering plants [3,4]. It has been reported that SBP genes play critical roles in regulating flowering, fruit ripening, phase transition, and other physiological processes. In Arabidopsis thaliana, AtSPL3, AtSPL4, and AtSPL5 are direct upstream activators of LEAFY, FRUITFULL, and APETALA1, and they redundantly promote flowering [5]. They also integrate developmental aging and photoperiodic signals in a process that involves the flowering locus T (FT)-flowering locus D (FD) module in A. thaliana [6]. In addition, AtSPL9 and AtSPL15 as well as AtSPL2, AtSPL10, and AtSPL11 are regarded as regulators of plastochron and branching [7,8]. AtSPL1 and AtSPL12 have been reported to play roles in plant thermotolerance during the reproductive stage [9]. AtSPL7 is a regulator of copper homeostasis and responses to light and copper [10]. There are also reports on SBP genes of other species: an SBP gene in Solanum lycopersicum (tomato) is critical for normal ripening [11]; OsSPL16 of Oryza sativa (rice) is a regulator of grain size, shape, and quality [12]; and OsSPL14 plays a role in controlling tiller growth in rice [13].
SBP genes encode a class of proteins that have a conserved DNA-binding domain (SBP-specific domain) that contains about 75 amino acid residues (aa). The SBPspecific domain is sufficient to bind to the GTAC core motif [2,[14][15][16]. There are three common structures in all SBP-specific domains: two zinc fingers and a nuclear localization signal (NLS). The NLS and the second zinc finger partly overlap [16]. Additionally, some SBP genes can be regulated by miRNAs (about 22-24 nt), which reduce protein levels at the transcriptional or translational stage by complementarily binding to their target mRNAs [17][18][19]. MiR156 plays the most important regulatory roles out of almost all the miRNAs that regulate SBP genes (with target sites located either in the coding region [CDS] or 3′ untranslated region [UTR]) [20,21]. It has been predicted that 10 of the 16 AtSPL genes are potential targets of miR156/157 (collectively known as miR156). Due to regulation by miRNAs, some SBP genes are involved in complex regulatory processes. For example, miR156 improves the drought tolerance of Medicago sativa by silencing SPL13 [22] and it regulates the juvenile-to-adult phase transition by regulating downstream target SBP genes [5,6,23]. Additionally, via miR156 regulation, AtSPL3 temporally regulates shoot development in A. thaliana [24].
Euphorbiaceae is a large and widespread plant family that consists of more than 8000 species, including herbs, perennial shrubs, and trees. They are evolutionarily diverse, and have various traits that allow them to adapt to dynamic environmental conditions. With the increasing demand for food, industrial raw materials, ornamental plants, and herbal medicines, Euphorbiaceae plants have become increasingly attractive. There are many agrieconomically important Euphorbiaceae species that have been widely cultivated, such as Ricinus communis (castor bean), Manihot esculenta (cassava), Jatropha curcas (physic nut), and Hevea brasiliensis (rubber tree). Castor bean can be cultivated at a large range of latitudes, and its oil is an important industrial raw material for producing lubricants and paints [25,26]. Cassava has a starchenriched root, and it has been a crucial food crop and is also ideal for bioethanol production [27,28]. Physic nut has seeds with a high oil content that can be processed into biodiesel [29,30]. The rubber tree is the most important source of natural rubber production, which is indispensable in daily life [31]. However, there are few studies on these non-model plants. More in-depth research, such as understanding the structure, evolution, and function of key gene families, is required to improve crop productivity and commercialization.
The SBP-box gene family has been identified and characterized in different plant species, such as A. thaliana [14], Malus domesrica (apple) [32], Physcomitrella patens (a moss species) [4], and Zea mays (maize) [33]. However, the SBP genes in Euphorbiaceae, and their evolutionary and functional characteristics, are rarely studied. Fortunately, the continuous publication of genome sequencing data [34][35][36][37] allows more in-depth research to be conducted on the Euphorbiaceae SBP-box gene family. Herein, we performed a genome-wide investigation of the SBP-box gene family in four Euphorbiaceae species. 77 SBP genes were identified using both local protein-protein Basic Local Alignment Search Tool (BLASTP) and hidden Markov model (HMM) searches. These genes were divided into three length ranges, and into 10 welldefined groups based on total sequence similarity and structural conservation. Duplication events and synteny blocks also supported our grouping scheme and revealed the details of the expansion process of Euphorbiaceae SBP genes. Additionally, a large amount of Euphorbiaceae SBP genes can be regulated by miR156. According to the expression profiles associated with different tissues and stress treatments, a large amount of miR-regulated SBP genes are highly differentially expressed in different tissues and the stress responses are ubiquitous among either miR-regulated or non-regulated SBP genes. Thus, we conducted a comprehensive analysis of Euphorbiaceae SBP genes, and provided valuable evolutionary information for further research.

Identification and characterization
Previous studies on the SBP-box gene family have mainly focused on the model plant A. thaliana. There are few studies on non-model plants such as Euphorbiaceae plants. Zhang and Ling reported on the identification and structural analysis of castor bean SBP genes, but they provided little function prediction information [38]. Here, we performed a comparative analysis of SBP genes from four representative Euphorbiaceae species: cassava, rubber tree, physic nut, and castor bean (Table 1). We systematically identified and characterized the SBP genes of Euphorbiaceae, and predicted their potential functions.
To comprehensively identify the SBP genes of each Euphorbiaceae species, we performed a whole-genome scan to identify protein-coding genes containing the SBPspecific domain by using both BLASTP and HMM search, and we then removed the proteins with incomplete SBPspecific domains. A total of 77 SBP genes containing 145 transcripts were identified (Additional file 1: Table S1). For each Euphorbiaceae species, the number of SBP genes varied from 15 to 26, comprising 15 in physic nut, 15 in castor bean, 21 in cassava, and 26 in rubber tree. The number of SBP genes was closely associated with genome size. For example, rubber tree and cassava had a relatively large number of SBP genes and they both experienced a recent genome duplication event [34,39].
To further characterize the SBP proteins, the basic properties including protein length, isoelectric point value, and molecular weight were analyzed (Additional file 1: Table  S2). The Euphorbiaceae SBPs covered a large range of lengths (140-1074 aa). Notably, the lengths exhibited a trimodal distribution (Fig. 1, Additional file 1: Table S2). The

Phylogenetic analysis and classification
To better understand the functions and evolutionary trajectory of the Euphorbiaceae SBP genes, a phylogenetic analysis of the 77 Euphorbiaceae SBPs plus 16 A. thaliana SPLs was implemented (Fig. 2). We first constructed a neighbor-joining phylogenetic tree involving the 93 SBPs. (Fig. 2a). The SBPs were divided into 10 distinct groups according to the phylogenetic analysis, namely, g1, g2, g3, g4, g5, g6, g7, g8, g9, and g10. This phylogenetic relationship was further confirmed by the maximum likelihood analysis showing that each group was  Fig. 1 The distribution of three length ranges of SBPs. Y-axis represents protein length (aa); X-axis lists three length ranges supported by a bootstrap value > 60% (Fig. 2b). Nine groups (all except g6) contained A. thaliana SPLs, which is consistent with previous results [14,40]. In addition, for the groups containing AtSPL genes, the Euphorbiaceae SBP genes were often close together, while the A. thaliana SBP genes were also close together. The protein characteristics of each group are summarized in Table 2. The exon number in each group exhibited a uniform tendency that was consistent with protein length (Fig. 2a).
We also conducted multiple sequence alignment for the conserved SBP-specific domain, which contained approximately 75 aa. Due to high structural similarity, we selected only one SBP gene per species per group for better visualization. All SBP-specific domains contained two zinc finger motifs and one nuclear localization signal (NLS) motif (Fig. 3). Nevertheless, the first zinc finger motif for g2 (Cys-Cys-Cys-Cys) was different from that in the other groups (Cys-Cys-Cys-His). For all the members of the 10 groups, compared with the first zinc finger, there was no structural difference in the second zinc finger (which was typically Cys-Cys-His-Cys). Moreover, each group had its own sequence features. For example, the second amino acid residue in g9 was L, while the fifth amino acid residue was K in g4 and G in its sister group g5.

Gene structure and conserved motif analysis
We further examined the structures of all SBP genes, comprising 77 in Euphorbiaceae and 16 in A. thaliana (Fig. 4a). The structural patterns were similar within each group but distinct between any two groups. In addition, the intron lengths of AtSPL genes were shorter than those in Euphorbiaceae genes. To identify the structural similarities and differences in SBPs between groups, a conserved motif analysis was performed. A total of 15 conserved motifs, including the SBP-specific domain (motif1), were found ( Fig. 4b, Additional file 2: Fig. S1). The motif Fig. 2 The phylogenetic tree. The neighbor-joining tree (a) was created using the MEGA7.0 program (bootstrap value set at 1000). The maximum likelihood tree (b) was constructed by PAUP* program. All these SBP proteins were divided into 10 groups, respectively are: g1, g2, g3, g4, g5, g6, g7, g8, g9, g10. The SBP genes in a specific group were marked with a specific color. The bootstrap values were marked by percentage, '%' was omited. The intron number for each SBP gene was displayed in a black bar outmost (a) the proteins in g2/4/5 were rich in motifs, sharply contrasting with the proteins in g3, which had only one motif. Some motifs were conserved across groups of different length ranges. For example, motif15 was shared for each middle-sized group and long-sized g5. Some motifs were group-specific: motif9 and motif14 were unique to g10, which was different from other middle-sized groups that contained only 2-3 motifs. Moreover, g4 and g5 shared many motifs, while motif5/13/4 were g5-specific and motif6 was g4-specific. Among the long-sized groups, g2 exhibited many differences in motifs compared to g4 and g5. In addition, g5 always contained both Ankyrin (ANK) and transmembrane regions, and the g5 proteins may be involved in protein-protein interactions.

Chromosomal locations and gene duplication events
The chromosomal distribution of the Euphorbiaceae SBP genes throughout the four Euphorbiaceae genomes was plotted using MapInspect software. Because of the lack of chromosome-level assembly data for physic nut, castor bean, and rubber tree, we plotted their SBP gene distribution at the scaffold level instead of the chromosome level ( Fig. 5, Additional file 1: Table S3). Gene duplication events among the Euphorbiaceae SBP genes were also examined ( Fig. 5, Additional file 1: Table S4.1). MCScan searching combined with micro-fragment comparison was used to find accurate duplicate gene pairs. Based on these two methods, 26 segment duplications were found: 12 in cassava, 6 in rubber tree, 4 in physic nut, and 4 in castor bean (Additional files 1: Table S4.1). The rubber tree contained the largest number of SBP genes but a relatively low number of duplications. Imperfect sequencing data partly led to the incomplete linear relationship between the number of duplicate gene pairs and the genome size. Segment duplications made a greater contribution to the Euphorbiaceae SBP gene expansions than tandem duplications (Additional file 1: Table S4.2). Six tandem duplication gene pairs were identified (Fig. 5). Interestingly, each SBP gene in g6 had one tandem duplication gene in g1 (HbSBP19-HbSBP20, HbSBP24-HbSBP23, JcSBP15-JcSBP6, RcSBP14-RcSBP4, and MeSBP8-MeSBP9), which suggests that these tandem duplication SBP genes may result in functional differentiation. All the predicted segment duplications were found within group, and they support our grouping scheme well. To further understand the evolutionary constraints on the Euphorbiaceae SBP genes, synonymous (Ks) and nonsynonymous (Ka) substitutions per site and their ratio (Ka/Ks) were calculated for the segment duplication gene pairs to explore their roles in the expansionary processes of SBP genes. The time to a certain duplication event can be calculated using the Ks value, as synonymous mutations accumulate at a relatively constant rate over time. Some Ks values were < 1 (marked -S) while others were 1-3 (marked -L) (Fig. 6). The bimodal distribution of the Ks values indicates that there were two large-scale duplication events. Ks-S duplications only existed in cassava and rubber tree, whereas Ks-L duplications were shared by all four Euphorbiaceae species (Additional file 1: Table S4.1). Given the Ks-L values in rubber tree, the -L duplications are likely to be associated with the triplication event related to all core eudicots [41]. The -L duplications generated branches consisting of conserved Euphorbiaceae genes. All the Ka-L values were greater than the Ka-S values (Fig. 6). However, the Ka-L/ Ks-L values were lower than the Ka-S/Ks-S ones, which mean that selection pressure on Ka was higher than Ks for SBP genes (Fig. 6). All Ka/Ks values were < 0.5 (Fig. 6), suggesting that the Euphorbiaceae SBP-box gene family underwent strong purifying selection to reduce detrimental mutations after duplication.

Synteny analysis
To explore the evolutionary process of the Euphorbiaceae SBP-box gene family, we conducted a comparative analysis of synteny blocks of genomes among the four Euphorbiaceae species and A. thaliana (Additional file 3: Fig. S2). Here, 141 syntenic blocks between Euphorbiaceae species were discovered (Additional file 3: Fig. S2). A high level of synteny relationships were found at both the species level (21/21 SBP genes in cassava, 15/15 in physic nut, 13/15 in castor bean, and 17/26 in rubber tree) and group level (all 10 groups were covered). Moreover, no intergroup synteny blocks were found (Additional file 1: Table S5), which is in accordance with the segment duplication results and validated our grouping scheme.

Prediction of microRNA target sites
We found the target sites of miR156 either in the CDS or 3'UTR (Table 3). For both A. thaliana and Euphorbiaceae, there was a similar ratio (2/1) of with-to withouttarget SBP genes. Long-sized SBP genes had no target sites, while both the middle-and short-sized SBP genes had target sites located either in CDS or 3'UTR (Table 2). However, one exception was that g1, a middle-sized group, contained no miR156 target (neither in A. thaliana nor in the Euphorbiaceae species).

Tissue expression profiles of JcSBP genes
To further illustrate the potential functions of each SBP gene, we conducted a comparative analysis of the expression data (from stem, inflorescence, buds, leaf, root,  Table S3. SBP gene pairs from segmental duplications are linked by blue lines; tandem duplications are marked by black circle. Each species are plotted in a unique part of (a) rubber tree, (b) cassava, (c) physic nut, (d) castor bean and seed) of physic nut and A. thaliana (Fig. 7). Because of the high similarity of SBP genes among the four Euphorbiaceae species, the analysis of the SBP genes of physic nut is very representative. Hierarchical clustering was used to visualize the global expression profile of the JcSBP genes (Fig. 7b). The expression patterns of the JcSBP genes could be divided into low differential expression between tissues (JcSBP4, JcSBP9, JcSBP11, JcSBP12, JcSBP10, JcSBP7, and JcSBP15) and high differential expression between tissues (JcSBP5, JcSBP13, JcSBP2, JcSBP6, JcSBP1, JcSBP3, JcSBP14, and JcSBP8). The former could be further divided into low expression genes (JcSBP10, JcSBP7, and JcSBP15) and high expression genes (JcSBP4, JcSBP9, JcSBP11, and JcSBP12).
There were significant differences in the expression profiles of JcSBP genes between the with-and withouttarget genes (Fig. 7b). The JcSBP genes of g2/4/5 (longsized groups) contained no target sites, and they were highly expressed without differential expression between tissues. In contrast, the with-target JcSBP genes in the middle-sized groups were highly differentially expressed in different tissues (with high expression in the buds and inflorescences, though several genes also played roles in the stem, leaf, or root). However, the tissue expression differences of the other with-target JcSBP genes (in the short-sized groups) were not as significant as the withtarget JcSBP genes in the middle-sized groups.
The expression patterns of AtSPL genes in g3 and g10 were significantly different from those in physic nut (Fig. 7). Regarding g3, the relative expression intensity of AtSPL genes was higher than those in physic nut, and they were highly expressed in more tissues. In contrast, regarding g10, the relative expression intensity of JcSBP genes was higher than AtSPL genes. The expression signal of AtSPL6 was barely observable. However, JcSBP2 and 13 were redundantly expressed in the stem, inflorescence, and root.

Stress response expression profiles of JcSBP genes
To further explore the possible physiological processes in which Euphorbiaceae SBP genes participate, the expression levels in physic nut in response to various abiotic stresses (salt, drought, and waterlogging) and hormonal treatments (gibberellin 3 [GA3], 6-benzylaminopurine [BA], and cytokinin) were obtained. Log2 transformations of the ratio of the treatment group data to their corresponding control group data are displayed in Fig. 8; log2 transformed values > 1 or < − 1 were viewed as representing differential expression.

Discussion
In view of their excellent agricultural traits, several Euphorbiaceae species have become important food sources or industrial raw materials. Cassava [27,28], physic nut [29,30], castor bean [25,42], and rubber tree [31] have been widely domesticated and cultivated. The continuously increasing quantity of genome sequencing data, genetic linkage maps, and abundance of high-throughput transcriptome sequencing data make further exploration of gene functions in non-model plants like Euphorbiaceae species possible. Previous studies on SBP genes have revealed their crucial roles in plant development, especially in flower development, signal transduction, and defense processes [5][6][7][8][9][10]. However, the functions of Euphorbiaceae SBPs are still unknown. In this study, genome-wide analyses (including the analyses of the evolutionary trajectory, miR156 regulation, and expression profiles) of the  Fig. 7 The tissue expression profiles. The tissue expression profiles of A. thaliana (a). Expression profiles of physic nut SBP genes among different tissues and development stages (b). The low expression differential groups were highlighted in blue (marked with L), and the high expression differential groups were highlighted in orange (marked with H). The blue groups can be further divided into high expressional and low expressional groups that marked with L-H and L-L respectively Euphorbiaceae SBP-box gene family were conducted to shed new light on Euphorbiaceae SBP genes. The phylogenetic relationships, synteny analysis, and tissue expression profiles showed that the SBP genes of Euphorbiaceae and A. thaliana are similar in structure, evolutionary trajectory, and functions. In light of the high similarity between SBP genes of Euphorbiaceae and A. thaliana, we can predict the functions of some of the SBP genes of Euphorbiaceae based on the wellstudied AtSPL genes. Regarding the long-sized groups, AtSPL7 (in g2) has been reported to be related to Cu homeostasis in A. thaliana, and it regulates the expression of Cu-responsive genes and is considered to be a central regulator of copper homeostasis [43,44]. The gene that is homologous to AtSPL7 was conserved in Euphorbiaceae and, similar to A. thaliana, it exhibited significantly high expression in the roots. Mutations of AtSPL14 (in g4) result in resistance to the fungal toxin fumonisin B1 [45]. AtSPL1 and AtSPL12 (in g5) play redundant roles in thermotolerance at the reproductive stage [9].
Regarding the middle-sized groups (g1/6/7/8/9/10), one of their remarkable characteristics is that they can be regulated by miR156 (all except g1). Due to regulation by miR156, these SBP genes play critical roles in plant development. AtSPL13 (in g9) has been implicated in delaying leaf outgrowth during germination [46]. AtSPL2, AtSPL10, and AtSPL11 (in g7) affect the morphological features associated with phase change [7]. AtSPL9 and AtSPL15 (in g8) play redundant roles in reproductive transition and vegetative phase change [8,47]. AtSPL8 (in g1) is related to seed formation, root development, and petal trichome [48,49]. As in A. thaliana, all the middle-sized JcSBP genes were differentially expressed between different tissues and exhibited high intensity expression, which suggests that they may be involved in different physiological processes and play critical roles in plant development and reproduction.
As we know, A. thaliana is monoecious, while physic nut is diecious; A. thaliana is a kind of biennial herb, while physic nut is a kind of perennial woody plant. It is Fig. 8 Expression profiles of physic nut SBP genes in response to abiotic stress and hormone stress treatments. The numerical values in different color scales represent log2 transform of the ratio of the experimental group and control group in a specific treatment condition worth exploring the functions of Euphorbiaceae SBP genes regarding the flowering process, phase transformation, seed development, etc. We found that the expression patterns of the SBP genes in g3 were significantly different between A. thaliana and physic nut, and there may be functional differences between them. In addition, regarding g10, the tissue expression profiles of A. thaliana were significantly different from those of physic nut in both relative expression intensity and the differential expression between different tissues. Moreover, g6 was absent from A. thaliana but conserved in Euphorbiaceae, and it was highly expressed in seeds and exhibited a relatively high response to salt, drought, and cytokinin. These results suggest that there may be some new functions or regulatory forms of SBP genes in Euphorbiaceae, and understanding these genes is helpful to further reveal the physiological regulation processes in Euphorbiaceae.
Sometimes plants are cultivated for their roots, sometimes for their seeds, and sometimes for their fruits. The formation of different tissues and organs may be related to different regulatory processes. Our study suggests that some SBP genes are differentially expressed in different tissues and organs, and may be associated with specific physiological processes. For example, physic nut and castor bean are cultivated for their seeds, so flower development and seed formation are important for a higher crop production. Both middle-and small-sized SBP genes are related to inflorescence or bud development according to their tissue expression profiles (Fig. 7b). In addition, several SBP genes were found to be related to seed development, such as JcSBP5/13/1/8, which express relative high in seeds (Fig. 7b). On the other hand, unlike physic nut and castor bean, cassava is cultivated for its roots, and JcSBP5/13 are highly expressed in the roots (Fig. 7b). Therefore, increasing the study of these SBP genes may contribute to the deeper understanding of specific physiological processes and subsequent agricultural genetics studies.
Conclusions SBP-box genes encode a series of plant-specific TFs, which have been identified and characterized in several species. Significant progress has been achieved regarding the identification of the functions of some SBP genes in several species, but little attention has been paid to nonmodel plants. In the present study, we identified 77 putative SBP genes in the genomes of four Euphorbiaceae species. From the results of the phylogeny analysis, we divided the Euphorbiaceae SBP genes into 10 independent groups, and the subsequent results regarding the structural analysis and the distribution of duplication gene pairs supported our grouping scheme. The genome comparison indicated that segment duplication played crucial roles in Euphorbiaceae SBP gene expansion, and all the duplication gene pairs were subjected to purify selection. In addition, two-thirds of Euphorbiaceae SBP genes may be regulated by miR156, and these miRregulated genes all belonged to the middle-or shortsized groups. Comparative synteny analysis between the genomes of five species (including A. thaliana) showed that a large number of SBP genes were located in syntenic regions, implying that these SBP genes probably come from common ancestors. Furthermore, to illustrate the probable functions of these SBP genes, we conducted a comparative analysis of the expression profiles of JcSBP and AtSPL genes in various tissues/organs. Most miRregulated JcSBP genes were more differentially expressed than miR-nonregulated JcSBP genes. G6 is conserved in Euphorbiaceae but not in A. thaliana, and we assume that it is functionally active as it was highly expressed in the buds and stems. However, the short-sized JcSBP genes were not as active as their homologous AtSPL genes, indicating there may be some functional differences between A. thaliana and Euphorbiaceae. Lastly, many JcSBP genes were up-or down-regulated in response to certain abiotic or phytohormone stresses, implying that they may be involved in the responses to various stresses or in physic nut development. Our data provide valuable information for further functional studies of Euphorbiaceae SBP genes. The flowering mechanism between A. thaliana and Euphorbiaceae and the high demand for increases in crop yield make the exploration of Euphorbiaceae SBP genes highly valuable.

Identification, characterization, and phylogenetic analysis
Both HMM [50] and BLASTP [51] searches were performed to accurately identify the SBP TFs in the Euphorbiaceae species. The well-characterized A. thaliana SBP protein sequences were used as queries for BLASTP searches (e-value ≤1e-10). The SBP-specific HMM profile (PF03110) was used for queries, and the HMMER toolkit was used in the HMM searches. The conserved SBP-specific domain was confirmed using Simple Modular Architecture Research Tool (SMART) [52] (http://smart.embl-heidelberg.de/), and the incomplete SBP-specific domains were discarded. In the cases involving multiple transcripts of the same gene, a dot followed by a serial number was added at the end of each name. The physicochemical properties, including protein length, molecular weight (MW), and isoelectric point (Pi), for the identified SBP proteins were predicted using the ExPASy Proteomics Server (https://prosite. expasy.org/) [53]. Multiple sequence alignment of SBP protein sequences was performed by Multiple Sequence Comparison by Log-Expectation (MUSCLE) in MEGA v7.0 [54]. A neighbor-joining tree was constructed using MEGA v7.0. The maximum likelihood tree was generated using the PAUP* program, employing the JTT substitution model and 100 bootstrap replicates [55].

Conserved motifs and gene structure analysis
The online Multiple Expectation Maximization for Motif Elucidation (MEME) toolkit was used to identify additional motifs (http://meme-suite.org/) [56], which were conserved and located outside the SBP-specific domain region. All SBP protein sequences were used for the queries. The parameters were set as follows: minimum width was 6, maximum width was 150, motif number was 15, and minimum number of sites was 2. Both SBP gene sequences and the corresponding coding sequences were uploaded to the online Gene Structure Display Server (GSDS v2.0; http://gsds.cbi.pku.edu.cn/) to obtain intron/exon structure information [57].

Chromosomal localization
A gene location map for each Euphorbiaceae species based on the chromosomal position of each SBP gene was generated by MapInspect (https://mapinspect.software.in former.com/). SBP gene locations of cassava were mapped into chromosomes, and SBP gene locations of the other three species were mapped into scaffolds due to their incomplete genome assembly information.

Detection of gene duplication events and synteny relationships
Duplicated gene pairs derived from tandem or segmental duplication were identified according to the method described in the Plant Genome Duplication Database [58]. An all-against-all BLASTP comparison (e-value ≤1e-10) provided gene pairs for syntenic clustering using MCScan v1.1 (e-value ≤1e-10) [59]. Segment duplication was also predicted by the micro-fragment comparison method. The SBP duplicate gene pairs from the above analysis were again examined by BLASTP (e-value ≤1e-10), and all the SBP genes obtained from the above analysis were used as anchors of microfragments generated by the collection of 20 upstream and 20 downstream coding genes. Tandem duplications were identified if two SBP genes were next to each other or they had one unrelated gene between them [60].
Estimation of synonymous (Ks) and nonsynonymous (Ka) substitutions per site and their ratio (Ka/Ks) SBP gene pairs caused by segmental duplication were used to estimate Ka, Ks, and their ratio. Coding sequences from segmentally duplicated SBP gene pairs were aligned using webPRANK (https://www.ebi.ac.uk/ goldman-srv/webprank/) [61]. KaKs_Calculator v2.0 [62] was used to compute Ka, Ks, and Ka/Ks. All the counting processes followed the YN model [63] (a simple model of voting). The Ka/Ks value can reveal the selective pressure of duplicated genes [64], and the Ks value can reflect the divergence time for duplication events. All-against-all BLASTP searches (e-value ≤1e-10) were conducted to investigate the synteny relationships of the proteomes of the four Euphorbiaceae species and A. thaliana. The synteny blocks were then calculated using MCScan v1.1 [59], and the synteny relationships were visualized using Circos v0.69-5 [65].

MicroRNA target prediction
MiR156 and miR157 were combined into the miR156 family in miRBase (https://www.mirbase.org/) [66], due to their highly similar structures. The well-characterized miR156 mature sequences from miRBase were set as the background data to search against the mRNA sequences of Euphorbiaceae SBP genes using psRNATarget program (http://plantgrn.noble.org/v1_psRNATarget/) [67] with default parameters. The detailed positions of miRNA (located in the CDS or 3'UTR region) were further determined on the basis of the locations of target sites and the CDS length.