The Ubiquitin-Conjugating Enzyme Gene Family in Longan (Dimocarpus longan Lour.): Genome-Wide Identification and Gene Expression during Flower Induction and Abiotic Stress Responses

Ubiquitin-conjugating enzymes (E2s or UBC enzymes) play vital roles in plant development and combat various biotic and abiotic stresses. Longan (Dimocarpus longan Lour.) is an important fruit tree in the subtropical region of Southeast Asia and Australia; however the characteristics of the UBC gene family in longan remain unknown. In this study, 40 D. longan UBC genes (DlUBCs), which were classified into 15 groups, were identified in the longan genome. An RNA-seq based analysis showed that DlUBCs showed distinct expression in nine longan tissues. Genome-wide RNA-seq and qRT-PCR based gene expression analysis revealed that 11 DlUBCs were up- or down-regualted in the cultivar “Sijimi” (SJ), suggesting that these genes may be important for flower induction. Finally, qRT-PCR analysis showed that the mRNA levels of 13 DlUBCs under SA (salicylic acid) treatment, seven under methyl jasmonate (MeJA) treatment, 27 under heat treatment, and 16 under cold treatment were up- or down-regulated, respectively. These results indicated that the DlUBCs may play important roles in responses to abiotic stresses. Taken together, our results provide a comprehensive insight into the organization, phylogeny, and expression patterns of the longan UBC genes, and therefore contribute to the greater understanding of their biological roles in longan.


Introduction
Ubiquitylation (also called ubiquitinylation or ubiquitination) is the covalent attachment of ubiqutin (Ub) to substrate proteins. Ubiquitin is a small protein containing 76 amino acids that is highly conserved in eukaryotes; only three residues differ between yeast, human, and plant species [1,2]. The process of protein ubiquitination (ubiquitin-proteasome system, UPS) is mediated through the action of three enzymes, E1 (Ub-activating enzyme, UBA), E2 (Ub-conjugating enzyme, UBC), and E3 (Ub ligase) [3]. Ub is first linked to E1 through an ATP-dependent reaction that creates a thioester bond between the C-terminus of Ub and the cysteine in the active site of E1. The activated Ub is then transferred via a thioester bond from E1 to a cysteine residue of E2, before ubiquitin is finally transferred either to a substrate directly aided by E3 or to a cysteine of an alternative ubiquitin protein ligase (E3s) by a second transthiolation reaction to the target substrate. Finally, the target proteins production being in Southeast Asia and Australia [35]. Common longan varieties, such as "SX", one of the main varieties in China, exhibit the "seasonal flowering" (SF) habit; floral bud induction requires a period of low temperature, and only the terminal meristem differentiates into an inflorescence. In order to obtain a stable high yield, off-season flowering in longan is achieved by chemical treatment with potassium chlorate (KClO 3 ) [36,37]. However, the induction effect varies in different regions and tree varieties. Therefore, the difficulty in flowering of these trees is a considerable problem in the longan industry. Therefore, the study of the molecular regulatory mechanisms of flower induction in longan is particularly important for understanding and solving the problems associated with flowering. One cultivar of longan, "SJ", flowers and bears fruits throughout the year under both high and low temperature, exhibit the "perpetual flowering" (PF) habit. This cultivar does not require a controlled environment; hence, it is a good model for studying longan flowering.
In the present study, we performed a genome-wide identification of UBC proteins in longan and analyzed their gene structures, conserved motifs, cis-elements and expression patterns in nine different tissues. This study also determined the expression profiles of longan UBC genes (DlUBC) during the three flowering stages in two longan cultivars, and measured their transcript abundance in response to different phytohormone treatments and abiotic stresses. This study provides a basis for future studies on the evolution and functions of the DlUBC gene family.

Phylogenetic Analysis of the DlUBC Genes
To categorize and investigate the evolutionary relationships of DlUBC genes, we constructed a phylogenetic tree by aligning the full-length UBC protein sequences for members of Saccharomyces cerevisiae (15), A. thaliana (48), and D. longan (40) (Figure 2). As shown in Figure 2, the results of our phylogenetic analysis revealed that all of the 103 UBC proteins could be categorized into 15 groups, and one group which doesn't contain any DlUBC based on >46% bootstrap support. The groups UBC9 and UBC12 functioned in SUMO and RUB1 conjugation pathways, respectively, and three UEV groups which lack the Cys active site. These groups were designated as UBC1, UBC2, UBC3/7, UBC4/5, UBC6, UBC8, UBC9, UBC10, UBC11, UBC12, UBC13, UBC14, UBC15, UEV1, UEV2 and UEV3. The 40 UBC members of longan were further divided into 15 groups that only contained two UEV group (UEV1 and UEV3). Interestingly, the groups UBC14, UBC15 and UEV3 were absent from the yeast genome, indicating that these groups may be plant-specific or were lost from the most common shared ancestor of yeast and plants. Additionally, UBC4 and UBC5 and UBC3 and UBC7, shared high homology; therefore, these groups were clustered as groups UBC4/5 and UBC3/7 in our study.

The Gene Structure and Motif Composition of the Longan UBC Gene Family
To further understand the similarity and diversity of motif composition among different DlUBCs, a neighbor-joining (NJ) phylogenetic tree was constructed using all full-length UBC protein sequences from longan. Using the yeast and Arabidopsis UBC proteins as references for classification, we subdivided the 40 UBC members from longan into 15 groups according to sequence similarity and topology ( Figure 1a).
According to the presence of the UBC domain and the N-or C-terminal extensions that are typically responsible for the functional differences between E2s, the E2s are divided into four types [4]. In the present study, 18 DlUBC proteins belong to Class I E2s; four, eleven and seven DlUBC proteins belong to Classes II, III, and IV E2s, respectively ( Figure 1b). The exon/intron structure analysis for the 40 longan UBC genes indicated that most of the coding sequences were disrupted by introns, except for the three genes DlUBC14, DlUBC30 and DlUBC37 (Figure 1c). The number of introns in the DlUBC genes ranged from zero to eight, with approximately 55% of the DlUBC genes possessing three or four introns. Phylogenetic analysis of DlUBC genes showed that most of the genes that clustered into the same group exhibited similar exon/intron structures. For example, all the members of groups UBC9 and UBC12 contained four introns (Figure 1c).
The longan UBC protein sequences were subjected to MEME (Multiple EM for Motif Elicitation); a total of 15 distinct motifs were identified and were designated as motif 1 to 15. The details of the conserved amino acid sequences and their lengths are shown in Table S3. The most common motif at the N-terminal was motif 4 (HPNIYSNGSICLDIL), which was found in 34 out of 40 (85%) longan UBCs, and motif 1 was also common at the N-terminal (77.5%). Most members in the same group shared similar motifs, and high variance was observed between the different groups ( Figure S1). The results also showed that some motifs were only found in one or two groups of DlUBC proteins. For example, motif 12 and 14 were found exclusively in groups UBC14 and UBC17, and motif 11 was only present in group UBC 4/5.

Phylogenetic Analysis of the DlUBC Genes
To categorize and investigate the evolutionary relationships of DlUBC genes, we constructed a phylogenetic tree by aligning the full-length UBC protein sequences for members of Saccharomyces cerevisiae (15), A. thaliana (48), and D. longan (40) ( Figure 2). As shown in Figure 2, the results of our phylogenetic analysis revealed that all of the 103 UBC proteins could be categorized into 15 groups, and one group which doesn't contain any DlUBC based on >46% bootstrap support. The groups UBC9 and UBC12 functioned in SUMO and RUB1 conjugation pathways, respectively, and three UEV groups which lack the Cys active site. These groups were designated as UBC1, UBC2, UBC3/7, UBC4/5, UBC6, UBC8, UBC9, UBC10, UBC11, UBC12, UBC13, UBC14, UBC15, UEV1, UEV2 and UEV3. The 40 UBC members of longan were further divided into 15 groups that only contained two UEV group (UEV1 and UEV3). Interestingly, the groups UBC14, UBC15 and UEV3 were absent from the yeast genome, indicating that these groups may be plant-specific or were lost from the most common shared ancestor of yeast and plants. Additionally, UBC4 and UBC5 and UBC3 and UBC7, shared high homology; therefore, these groups were clustered as groups UBC4/5 and UBC3/7 in our study.

Tissue-Specific Expression Patterns of DlUBC in Longan
To assess the potential functions of DlUBC genes during longan development, the expression profiles of 40 DlUBC genes in root, stem, leaf, seed, young fruit, pulp, pericarp, flower, and flower bud were investigated by RNA-seq analysis. The RNA-seq data for 40 DlUBC genes (Table S4) was downloaded from the NCBI database and a heat map of their expression was generated ( Figure 3). Results showed that almost all DlUBCs were expressed in flowers and flower bud except DlUBC15

Tissue-Specific Expression Patterns of DlUBC in Longan
To assess the potential functions of DlUBC genes during longan development, the expression profiles of 40 DlUBC genes in root, stem, leaf, seed, young fruit, pulp, pericarp, flower, and flower bud were investigated by RNA-seq analysis. The RNA-seq data for 40 DlUBC genes (Table S4) was downloaded from the NCBI database and a heat map of their expression was generated ( Figure 3). Results showed that almost all DlUBCs were expressed in flowers and flower bud except DlUBC15 and DlUBC35. Furthermore, 92.5% of the DlUBCs were expressed in the pericarp, root, stem and young fruit; and 90% were expressed in the leaf, pulp and seed. Approximately 87.5% (35 of 40) of the DlUBC genes were expressed in each tested tissue. A total of three DlUBC genes (DlUBC10, 33, and 37) had low expression in all tested tissues. Furthermore, DlUBC15 and DlUBC35 could not be detected in all tested longan tissues and DlUBC17 only displayed a significantly low expression in flowers. The DlUBC1 gene showed no expression in leaf, seed, and young fruit, and low expression in the remaining tissues (Table S4). It is worth noting that 14 genes (DlUBC2, 3, 5, 6, 9, 16, 18, 22, 23, 25, 28, 29, 34 and 38) were highly expressed in the nine longan tissues.

Comparative Expression Profiles of the Two Longan Species during Flowering
Flowering is a critical event in the life cycle of plants, especially in fruit trees. However, the mechanisms of flower induction in longan have not been elucidated. In the present study, we also analyzed the expression patterns of 40 DlUBC genes in two longan species during the three

Comparative Expression Profiles of the Two Longan Species during Flowering
Flowering is a critical event in the life cycle of plants, especially in fruit trees. However, the mechanisms of flower induction in longan have not been elucidated. In the present study, we also analyzed the expression patterns of 40 DlUBC genes in two longan species during the three flowering stages by RNA-seq analysis (Table S5). One heat map was constructed based on the log 10 (FPKM + 0.01) values for the 40 DlUBC genes (Figure 4). DlUBC genes differentially expressed during the three flowering stages of the two longan species were identified based on the criteria for p values < 0.05 and fold changes ≥ 2. Results showed that all the 40 DlUBC genes were constitutively expressed during the three flowering stages in the cultivar 'SX', while 11 DlUBC genes were differentially expressed in 'SJ'. Among these 11 DlUBC genes, seven (DlUBC1, 10, 13, 14, 19, 20 and 36) were up-regulated during the three flowering stages, and three genes (DlUBC21, 26 and 30) were down-regulated. Moreover, one gene (DlUBC37) was down-regulated in the first two stages and then upregulated in the third stage. transcript levels of all six DlUBC genes did not exhibit any significant differences in 'SX' longan between the three flowering stages. In addition, the relative expression level of DlUBC11, DlUBC16 and DlUBC31 did not exhibit any significant differences in 'SJ' during the three flowering stages. The expression levels of DlUBC19 and DlUBC20 were upregulated at the second and third stage. The transcript levels of DlUBC30 were downregulated at the second and third stage. In general, the expression levels obtained by qRT-PCR for these genes are similar to the results obtained from the RNA-seq data.  To validate the expression levels obtained from the RNA-seq data, six DlUBC genes (DlUBC11, 16, 19, 20, 30 and 31) were randomly selected from six different longan UBC groups for quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR) analysis ( Figure 5). The transcript levels of all six DlUBC genes did not exhibit any significant differences in 'SX' longan between the three flowering stages. In addition, the relative expression level of DlUBC11, DlUBC16 and DlUBC31 did not exhibit any significant differences in 'SJ' during the three flowering stages. The expression levels of DlUBC19 and DlUBC20 were upregulated at the second and third stage. The transcript levels of DlUBC30 were downregulated at the second and third stage. In general, the expression levels obtained by qRT-PCR for these genes are similar to the results obtained from the RNA-seq data.

Differential Regulation of DlUBCs in Response to Stress and Hormonal Treatments
Subsequently, the expression patterns of 40 DlUBC genes were investigated in response to hormonal and various stress using qRT-PCR ( Figure 6).

Differential Regulation of DlUBCs in Response to Stress and Hormonal Treatments
Subsequently, the expression patterns of 40 DlUBC genes were investigated in response to hormonal and various stress using qRT-PCR ( Figure 6).
In the 40 DlUBC genes, DlUBC6, 10, 16, 24, 26 and 32 showed no significant differential expression in response to the treatments. The remaining 34 DlUBC genes were up-regulated or down-regulated in at least one tested treatment. We identified 17 DlUBC genes with different expression levels under SA treatment, in which seven genes (DlUBC4, 9, 11, 17, 19,

Analysis Related Cis-Elements in the Candidate DlUBC Genes
To further analyze the potential roles of DlUBC genes in response to various responses, a 1.5 kb upstream regulatory region (promoter) of DlUBC genes were used to search for cis-elements. Of the 40 genes, 1.5 kb upstream regulatory region could be fetched in 39. Only six promoter bases of DlUBC40 could be fetched as only these many bases are available upstream in the assembled scaffold which it belongs. All of DlUBC genes shared the light-responsive boxes and

Analysis Related Cis-Elements in the Candidate DlUBC Genes
To further analyze the potential roles of DlUBC genes in response to various responses, a 1.5 kb upstream regulatory region (promoter) of DlUBC genes were used to search for cis-elements. Of the 40 genes, 1.5 kb upstream regulatory region could be fetched in 39. Only six promoter bases of DlUBC40 could be fetched as only these many bases are available upstream in the assembled scaffold which it belongs. All of DlUBC genes shared the light-responsive boxes and stress-responsive boxes in their promoter. Hormones-related cis-elements, such as MeJA, salicylic acid, gibberellin, auxin and ethylene, were existed in the promoter of all DlUBC genes except DlUBC29. Additionally, circadian-related cis-elements were found in the promoter of thirty-two DlUBC genes and Meristem-related cis-elements only presented in the promoter of seventeen DlUBC genes (Figure 7, Tables S6 and S7). These results indicated that DlUBC genes may be regulated by various cis-elements within the promoter during growth and stress responsive. stress-responsive boxes in their promoter. Hormones-related cis-elements, such as MeJA, salicylic acid, gibberellin, auxin and ethylene, were existed in the promoter of all DlUBC genes except DlUBC29. Additionally, circadian-related cis-elements were found in the promoter of thirty-two DlUBC genes and Meristem-related cis-elements only presented in the promoter of seventeen DlUBC genes (Figure 7, Tables S6 and S7). These results indicated that DlUBC genes may be regulated by various cis-elements within the promoter during growth and stress responsive.

Discussion
Ubiquitin-conjugating enzymes (E2s) have been characterized and analyzed in both prokaryotes and eukaryotes [9]. However, neither a genome-wide identification nor a comprehensive assessment of this gene family in longan has been previously reported. Recently, the successful sequencing of the longan genome has made it possible to analyze this gene family at the whole-genome level [38].

Discussion
Ubiquitin-conjugating enzymes (E2s) have been characterized and analyzed in both prokaryotes and eukaryotes [9]. However, neither a genome-wide identification nor a comprehensive assessment of this gene family in longan has been previously reported. Recently, the successful sequencing of the longan genome has made it possible to analyze this gene family at the whole-genome level [38].
The phylogenetic relationship analysis showed that all the 103 UBC proteins could be categorized into 16 groups. Eleven groups (including UBC1, 2, 3/7, 4/5, 6, 8, 9, 10, 11, 12 and 13) were present in S. cerevisiae, A. thaliana, and D. longan, as well as in rice, tomato, and maize [8,9,16,39,40]. This result suggests that these 11 groups may have evolved before the divergence of the ancestor of yeast and plants. The ubiquitin E2 enzyme variant (UEV) proteins are similar to E2s in both sequence and structure, but lack a catalytic cysteine residue, and thus are unable to form a thiol-ester linkage with ubiquitin [44]. To comprehensively understand the function of longan UBC proteins, the UEVs were also considered. In the present study, three UEV genes (DlUBC4, DlUBC6 and DlUBC9) were existed in the longan genome, which is fewer than in Arabidopsis, rice, and maize [8,39,40]. Additionally, the groups UBC14 and UBC15 are absent in the yeast genome, indicating that these groups may have been lost in the ancestor of yeast or have evolved after the divergence of the ancestor of yeast and plants. The number of UBC genes differed among the groups too. For instance, the UBC1, 2, 10 and 15 groups only contained one DlUBC gene each, while the largest group (UBC4/5) included eight genes. Similar results were found in other studies [7,8,41], suggesting group UBC4/5 might have more diverse functions than other groups. In addition, there are some minor differences in the topologies of the UBC genes in Arabidopsis among different studies. For example, AtUBC31, clustered into the UBC 4/5 group in the previous studies [7,8], was not placed in any groups in our study. These differences in protein classification could have resulted from different parameter settings or methods during the phylogenetic analyses. Accumulated data suggests that UBC genes play important roles in diverse plant development processes and have different expression patterns in different organs [8,45]. For example, in Arabidopsis, AtUBC1 and AtUBC2 are ubiquitously expressed in roots, leaves, flowers, and seedlings [34]. The double mutant of Arabidopsis UBC13A/B displays strong phenotypes, including shortened primary roots, a reduced number of lateral roots, and few and short root hairs [46,47]. In banana, MaUBC10, 11, 33, 34, and 61 are highly expressed in most organs, but especially in roots, stems, and leaves; while MaUBC6, 11, 34, 35, 45, and 61 were highly expressed in stems, implying that these genes were likely to be involved in basal metabolic or housekeeping functions in the banana development [7]. In papaya, all 34 CpUBC genes showed organ-specific expression patterns; nineteen (CpUBC1, 2, 3, 5, 6, 9, 10, 11, 12, 15, 17, 20, 23, 24, 26, 30, 31, 33 and 34) were highly expressed in male flowers and two genes (CpUBC21 and CpUBC22) were expressed in female flowers which suggests that these genes may be involved in the development of floral sex organs [41]. Consistent with the previous studies, in the present study, most DlUBC genes were expressed widely in the different organs that we examined, suggesting that DlUBC genes may be play diverse roles in longan organ development (Table S4 and Figure 3). For example, DlUBC3, belonging to group UBC2, is orthologous to AtUBC1 and AtUBC2, and has ubiquitous expression in roots, leaves, and flowers. Meanwhile, our results also showed that several genes showed a specific expression in longan organs. For instance, DlUBC17 was only weakly detected in flowers and DlUBC19 were higher expressed in roots, which indicated that these two genes might be involved in the development of flowers or roots, respectively. In general, these results indicate that DlUBC genes may play various roles in the development of different longan tissues.
Flowering is a transition from vegetative to reproductive development, and is one of the most important events in the life cycle of higher plants, because it is vital for reproductive success [18,48]. This transition is coordinated through a diverse array of signaling networks that integrate various endogenous and exogenous signals [23]. In past decades, we have gained increasing knowledge of flowering time regulation in model species such as Arabidopsis [20] and many family genes involved in this regulation have been identified, such as WRKYs and ASMT [29,49]. Although UBC proteins have important roles in plant growth and development, little is known about its functions in the process of flower induction. For example, the Arabidopsis UBC1 and UBC2, together with two closely related RING-type E3s called HUB1 (HISTONE MONOUBIQUITINATION1) and HUB2, are involved in histone 2B monoubiquitination and the regulation of flowering time [34,50]. For longan, several studies indicated that the homologues, such as SHORT VEGETATIVE PHASE (SVP), GIGANTEA (GI), F-BOX 1 (FKF1), EARLY FLOWERING 4 (ELF4), CO and FLC, might be involved in the control of flowering by using RNA-seq analysis [36,37]. However, to date, the role of UBC proteins in the flower induction in longan has not been previously studied. In the present study, the expression of 40 DlUBC genes was evaluated during three different flowering stages by using RNA-seq. Interestingly, the results showed that all of the 40 DlUBC genes were constitutively expressed in the three flowering stages in the "SX" longan variety, which flowers only once a year. Additionally, 11 DlUBC genes were differentially expressed in the "SJ" longan variety, which flowers throughout the year (Figure 4). Meanwhile, the expression levels measured by qRT-PCR for Six DlUBC genes (DlUBC11, 16, 19, 20, 30 and 31) randomly selected were similar to the results obtained from the RNA-seq data ( Figure 5). These results suggesting that those DlUBC genes may participate in flower induction, especially involved in the regulation of PF habit in longan. However, DlUBC3, which is orthologous to AtUBC1 and AtUBC2, did not show any change during the three flowering stages. This result is consistent with the expression of AtUBC3, the other member of group UBC 2, which does not show redundancy with AtUBC1 and AtUBC2. Furthermore, only the UBC1 UBC2 double mutant without UBCs has an early flowering phenotype [50]. We speculate that these orthologous genes may be involved in different signaling pathway in Arabidopsis and longan. In summary, we propose that these 11 DlUBC genes play crucial roles in longan flowering, and need further investigation.
Longan is frequently challenged by abiotic stressors such as high salinity, drought, and extreme temperatures. Recent studied have shown that UBC proteins are widely involved in signaling and response to these stresses in many species [45]. For example, three rice genes (OsUBC2, 5 and 18) and five Arabidopsis genes (AtUBC13, 17, 20, 26, and 31) in the UBC family were significantly down-regulated, whereas only three rice genes (OsUBC13, 15 and 45) were significantly up-regulated under salt and drought stresses [51]. In maize, 16, 20, and over half of the ZmUBC genes (48 genes) were significantly up-regulated under drought, cold, and salt conditions, respectively [8]. Consistent with previous studies, in the present study, the mRNA levels of 26 and 14 DlUBC genes were up-or down-regulated by heat or cold treatment, respectively ( Figure 6). These results suggest that those genes might play important roles under high or low temperature conditions. To date, several studies indicated that E3 proteins respond to hormonal treatment. For instance, the RING E3 ligases AIRP1 and AIRP2 are responsible for reducing root growth rate in response to ABA [52]. However, there are few studies on the interaction between UBC protein and hormones. In the present study, 17 and seven DlUBC genes had different expression levels during SA and MeJA treatments, respectively. These results indicate that these DlUBCs could potentially play vital roles in stress and hormone responses. Differential responses of some family genes are regulated by the presence of cis-elements in their promoter region [53][54][55]. For example, Morus013217 which contained three LTREs in its promoter regions showed a strong response to cold stress [53]. Similar results also found in our study. For instance, one HSE cis-element was found in the promoter regions of DlUBC9, which showed an induce response to heat stress. DlUBC9 and DlUBC22 showed responsiveness to SA treatment, and TCA-elements were found in their promoters (Figure 7 and Table S7). Thus, these cis-elements could provide more evidence of DlUBC genes in response to different stress or hormonal signaling.

Identification of the Longan Conjugating Enzyme Family Gene
Genome sequences of longan have recently become available and were downloaded from the NCBI Sequence Read Archive (SRA315202) or ftp://climb.genomics.cn/pub/10.5524/100001_101000/ 100276/ [38]. To identify potential members of the DlUBC gene family, the hidden Markov model (HMM) profile of the ubiquitin-conjugating enzyme domain (PF00179) was extracted from the Pfam database (http://pfam.xfam.org/family/PF00179) [56] and used to search for putative UBC proteins from the longan genome sequence with HMMER 3.0 (http://hmmer.janelia.org/). The default parameters were adopted, and the cutoff value was set to 0.01. Subsequently, BLAST searches using all Arabidopsis and Saccharomyces UBC protein sequences as queries were performed with default parameters. Finally, all candidate sequences were examined to confirm the presence of the conserved UBC domain (PF00179) using SMART (http://smart.emblheidelberg.de) and Pfam (http://pfam.xfam.org) database analyses [57].

Sequence Analysis
The molecular weight (MW), number of amino acids, open reading frame (ORF) length, and isoelectric point (pI) of DlUBCs were calculated using ExPASy online tools (http://expasy.org/ tools/) [58]. Gene Structure Display Server (GSDS) version 2.0 was used to display the intron and exon junctions and the arrangements of DlUBC genes [59]. The conserved motifs of DlUBC proteins were identified by MEME (http://meme.sdsc.edu/meme/cgi-bin/meme.cgi) [60] with the following optimized parameters: any number of repetitions, a maximum number of 15 motifs, and an optimum width of each motif between six and 50 residues.

Sequence Alignment, Cis-Elements in the Promoters and Phylogenetic Analysis
Sequences of 15 S. cerevisiae and 48 Arabidopsis UBC proteins were described previously [9,39] and obtained from the Saccharomyces Genome Database (http://www.yeastgenome.org/) and TAIR (http://www.arabidopsis.org/), respectively. The 1,500-bp sequences upstream of the transcription start site of candidate DlUBC genes were extracted from the longan genome sequences. PlantCARE software (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) was used for searching the cis-acting elements [61]. For phylogenetic analysis, the UBC protein sequences of S. cerevisiae, Arabidopsis, and longan were aligned using Clustal X 1.83 (http://www.bio-soft.net/fomat.html). Based on this alignment, a bootstrapped neighbor-joining (NJ) tree was constructed using MEGA version 6.0 (http://www.megasoftware.net) and a bootstrap test replicated 1000 times [62]. To assess the phylogenetic relationships among the members of the longan UBC gene family, a phylogenetic tree was constructed according to the alignment of only longan proteins. All DlUBC proteins were classified into groups based on their structural features and evolutionary relationships.

Expression Analysis of Longan UBCs in Various Tissues and at Different Flowering Stages
The RNA-seq data for the "SJ" variety was downloaded from the NCBI Sequence Read Archive (GSE84467) and used to analyze the expression patterns of UBC genes in the root, stem, leaf, seed, young fruit, pulp, pericarp, flower, and flower buds. Fragments per kilobase of exon model per million mapped values were log 10 -transformed, and heat maps with hierarchical clustering were designed using the software Mev 4.9.0 (http://tm4.org) [63].
Three pairs of nine-year-old "SJ" trees, which exhibit the perpetual flowering habit, and "SX" trees, which exhibit the seasonal flowering habit, were used in this study. Those trees were grown at experimental orchard of the South Subtropical Crops Research Institute of the Chinese Academy of Tropical Agricultural Science in Zhanjiang (110 • 16 E, 21 • 10 N), China. Three different kind apical buds from the dormant stage (before the emergence of floral primordial) (T1), the emergence of floral primordia stage (T2), and the floral organ formation stage (T3) of "SJ" and "SX" were identified by a histological analysis [64]. Samples of each stage of "SJ" are abbreviated SJT1, SJT2, SJT3, and samples of different development stages in "SX" are abbreviated SXT1, SXT2 and SXT3. The samples obtained for the SXT1, SXT2 and SXT3 were collected on 20 November 2016, 24 December 2016, and 1 January 2017, respectively. The three kind samples of "SJ" were obtained at the same time compared to "SX". For each sample, we used three biological replicates from three different trees. Each biological replicate contained mixed buds. All samples were collected from 10:00 to 12:00 a.m., and were frozen immediately in liquid nitrogen and stored at −80 • C. Total RNA were extracted separately from the bud samples of three biological replicates using the quick RNA Isolation Kit (Hua Yue Yang Bio Co., Ltd., Beijing, China) according to the manufacturer's instructions, and the genomic DNA residues were removed during RNA extraction. RNA concentration and quality were tested in an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA). RNA quality was also confirmed by RNase free agarose gel electrophoresis. RNA-seq libraries were constructed as previously described [61] and sequenced on the Illumina HiSeq™ 2000 platform (Illumina Inc., San Diego, CA, USA). Before assembly, adapter sequences were removed from the raw reads. Then low quality reads with over 50% bases with quality scores of 5 or lower and/or over 10% bases unknown (N bases) were removed from each dataset to gain more reliable results. After that, the clean reads of high quality from all the 18 samples were mapped to the longan genome databases [38], respectively. After alignments, raw counts for each D. longan transcript and each sample were derived and were normalized to Reads Per Kilobase of transcript per Million mapped reads (FPKM). Differentially expressed genes (fold changes > 2 and adjusted p-value < 0.05) were identified by the DESeq package. The RNA-seq data have been uploaded to the NCBI Sequence Read Archive (SRS2241241, SRS2241242, SRS2241243, SRS2241244, SRS2241245, SRS2241246, SRS2241247, SRS2241248, SRS2241249, SRS2241250, SRS2241251, SRS2241252, SRS2241253, SRS2241254, SRS2241255, SRS2241256, SRS2241257 and SRS2241258).

Hormonal and Stress Treatments and Expression Profiling Using qRT-PCR
In this study, 27 one-year-old uniform grafted seedlings of "SJ", obtained from the South Subtropical Crops Research Institute of the Chinese Academy of Tropical Agricultural Science in Zhanjiang (110 • 16 E, 21 • 10 N), were used for stress and hormonal treatments. For hormone treatments, three seedlings were treated with methyl jasmonate (MeJA) or SA solution (100 µM) for 4 h at 28 • C, respectively. Meanwhile, three seedlings sprayed with water were used as a control. For heat and cold stresses, three samples were grown at 42 • C or 0 • C for 4 h, respectively, and three samples grown at 28 • C were used as a control. All of the treatments were performed in a greenhouse. Six leaves were collected from each seedling and all samples were immediately frozen in liquid nitrogen and stored at −80 • C for expression analysis. Total RNA was extracted from leaves using the SuperFast new plants of RNA extraction kit while eliminate genome DNA following the manufacturer's instructions (Hua Yue Yang Bio Co., Ltd., Beijing, China). First-strand cDNA was synthesized by reverse transcription of total RNA (500 ng) using PrimeScript RTase (TaKaRa Biotechnology, Dalian, China). Gene-specific primers were designed according to the DlUBC gene sequences using Primer Premier 5.0 and checked using BLASTn in NCBI (Table S1). In addition, the longan Actin1 gene (Dlo_028674) was used as an internal control for normalization of the expression data. Real-time PCR was performed using a Bio-Rad real-time thermal cycling system (LightCycler 480; Bio-Rad Laboratories, Inc., Hercules, CA, USA) and SYBR-green to assess the expression levels of the candidate DlUBC genes. Each reaction consisted of 10 µL of 2 × SYBR Premix Ex Taq II (Takara Bio), 40 ng cDNA, and 250 nM of each primer in a final volume of 20 µL. The following PCR conditions were used: 94 • C for 15 min, followed by 40 cycles of 95 • C for 15 s, 58-63 • C for 20 s, and 72 • C for 30 s. The relative mRNA levels of the genes were measured using the cycle threshold (Ct) 2 (−∆Ct) method. The analysis included cDNA from three biological samples for each tissue, and all the reactions were run in triplicate. In the comparative expression analysis of DlUBC gene expression, genes that were up-or down-regulated at least two-fold were considered differentially expressed.

Statistical Analysis
Data were analyzed using variance (ANOVA) and the means were compared by the t test at the 5% level using the SPSS 11.5 software package (SPSS, Chicago, IL, USA).

Conclusions
A total of 40 putative DlUBC genes were identified in the longan genome and were grouped into 15 groups based on a phylogenetic analysis. The gene structure, conserved motifs, cis-elements and expression profiling, which may be related to their biological functions, were systematically analyzed. In each group, the exon-intron junctions and sequence motifs were highly conserved. The expression patterns of the DlUBC genes in various tissues showed that these genes might have important functions in longan growth and development. Based on our previous transcriptome data, we also analyzed the expression patterns of 40 DlUBC genes in two longan species during the three flowering stages. The results show that all the 40 DlUBC genes were constitutively expressed in all the three flowering stages in the "SX" longan variety, while 11 DlUBC genes were differentially expressed in the "SJ" longan variety. In addition, the expression levels obtained by qRT-PCR for six DlUBC selected genes (DlUBC11, 16, 19, 20, 30 and 31) were similar to the results obtained from the RNA-seq data. The expression results suggest that DlUBC genes may be involved in the regulation of flower induction. Furthermore, the expression patterns of DlUBC genes show that they play potentially important roles in mediating the effects of stress induced by SA, MeJA, and extreme temperatures. The results of our study establish a foundation for future studies on the functions of DlUBC genes in organ development and plant stress response, and for further elucidation of the potential functions of the DlUBC genes in longan varieties.