Comparative Genomic and Transcriptomic Analyses of Family-1 UDP Glycosyltransferase in Prunus Mume

Glycosylation mediated by Family-1 UDP-glycosyltransferases (UGTs) plays crucial roles in plant growth and adaptation to various stress conditions. Prunus mume is an ideal crop for analyzing flowering for its early spring flowering characteristics. Revealing the genomic and transcriptomic portfolio of the UGT family in P. mume, a species in which UGTs have not yet been investigated, is therefore important. In this study, 130 putative UGT genes were identified and phylogenetically clustered into 14 groups. These PmUGTs were distributed unevenly across eight chromosomes and 32 tandem duplication and 8 segmental duplication pairs were revealed. A highly conserved intron insertion event was revealed on the basis of intron/exon patterns within PmUGTs. According to RNA-seq data, these PmUGTs were specifically expressed in different tissues and during the bud dormancy process. In addition, we confirmed the differential expression of some representative genes in response to abscisic acid treatment. Our results will provide important information on the UGT family in P. mume that should aid further characterization of their biological roles in response to environmental stress.


Introduction
Increasing evidence is suggesting that glycosylation mediated by glycosyltransferases (GTs) plays crucial roles in plant growth and response to biotic and abiotic stresses [1]. According to numerous studies, GTs catalyze the transfer of sugar moieties from active sugar molecules to a variety of acceptor molecules, namely, hormones, lipids and some other small molecules [1,2]. The formation of a glycosidic bond can change an acceptor's chemical properties and bioactivity, adjustments that are essential for the maintenance of cellular homeostasis. In addition, conjugation by GTs allows plant cells to modulate their biochemical proprieties and thus have a strong influence on their biological activity and compartmental storage [3].

Identification of the Putative UGTs in P. mume
The UGT proteins play crucial roles in various plant developmental processes. Three strategies were used to identify candidate UGT genes in P. mume, as mentioned in the Material and Methods part. Subsequently, 130 potential UGT protein sequences were identified, which were named based on the chromosomal location of the corresponding genes. All these 130 putative UGT sequences started with a methionine and were full-length sequences. The protein length, molecular weight, isoelectric point and putative subcellular localization of these proteins varied widely (Supplementary Table S1). The protein sequence length and molecular weight were ranged from 279 (Pm027884) to 764 (Pm000189) amino acids (aa) and 31.25 (Pm027884) to 85.14 (Pm000189) kDa, with an average length of 470 aa and 52.45 KDa, respectively. The predicted isoelectric points varied from 4.63 (Pm000211) to 8.79 (Pm019106). Protein subcellular localization of 130 PmUGTs was also predicted by bioinformatics methods. Most of PmUGT proteins were predicted to be located in the chloroplast (75 members). 32 PmUGTs were predicted to be located in the cytoplasm and 13 were in the nucleus. More detailed information was provided in Supplementary Table S1.

Chromosomal Distribution, Duplication and Divergence
The genomic distribution of 130 PmUGTs revealed that 121 PmUGTs distributed across eight chromosomes and nine located on scaffolds ( Figure 1). There were 26 UGTs on chromosome 2, followed by 20, 19 and 17 members on chromosome 4, 1 and 6, respectively. Tandem and segmental duplication events were also analyzed for its importance to elucidate the chromosomal/gene segments and tandem exons. As shown in Figure 1, 32 gene pairs, including 56 PmUGTs, involved in tandem duplication. Moreover, eight gene pairs (Pm001086/Pm030144, Pm002233/Pm004391, Pm002464/Pm021221, Pm005787/Pm026572, Pm010818/Pm014833, Pm018404/Pm024073, Pm019826/Pm024073, Pm025006/Pm030552) were involved in the segmental duplication events (Figure 1, Supplementary Table S2). These results suggest that tandem duplication might play major roles in the PmUGT family amplification. When compared with P. persica genome, 23 segmental duplications pairs were found, as detailed in Table 1. To further analyze the syntenic relationships of UGTs between P. mume and P. persica, we mapped the 23 segmental duplication pairs to the duplicated blocks ( Figure 2).

Identification of the Putative UGTs in P. mume
The UGT proteins play crucial roles in various plant developmental processes. Three strategies were used to identify candidate UGT genes in P. mume, as mentioned in the Material and Methods part. Subsequently, 130 potential UGT protein sequences were identified, which were named based on the chromosomal location of the corresponding genes. All these 130 putative UGT sequences started with a methionine and were full-length sequences. The protein length, molecular weight, isoelectric point and putative subcellular localization of these proteins varied widely (Supplementary  Table S1). The protein sequence length and molecular weight were ranged from 279 (Pm027884) to 764 (Pm000189) amino acids (aa) and 31.25 (Pm027884) to 85.14 (Pm000189) kDa, with an average length of 470 aa and 52.45 KDa, respectively. The predicted isoelectric points varied from 4.63 (Pm000211) to 8.79 (Pm019106). Protein subcellular localization of 130 PmUGTs was also predicted by bioinformatics methods. Most of PmUGT proteins were predicted to be located in the chloroplast (75 members). 32 PmUGTs were predicted to be located in the cytoplasm and 13 were in the nucleus. More detailed information was provided in Supplementary Table S1.

Chromosomal Distribution, Duplication and Divergence
The genomic distribution of 130 PmUGTs revealed that 121 PmUGTs distributed across eight chromosomes and nine located on scaffolds ( Figure 1). There were 26 UGTs on chromosome 2, followed by 20, 19 and 17 members on chromosome 4, 1 and 6, respectively. Tandem and segmental duplication events were also analyzed for its importance to elucidate the chromosomal/gene segments and tandem exons. As shown in Figure 1 Table S2). These results suggest that tandem duplication might play major roles in the PmUGT family amplification. When compared with P. persica genome, 23 segmental duplications pairs were found, as detailed in Table 1. To further analyze the syntenic relationships of UGTs between P. mume and P. persica, we mapped the 23 segmental duplication pairs to the duplicated blocks ( Figure 2).    All the segmental duplicated UGT gene pairs had undergone a whole-genome duplication and the Ka/Ks ratios were less than 1. This result indicated that these UGTs experienced negative selection during species evolution process. Moreover, the divergence times of the duplicated UGTs at P. mume were significantly larger than that between P. mume and P. persica. The divergence time of the eight duplicated pairs at P. mume genome spanned from 47.84 (Pm025006-Pm030552) to 99.11 (Pm005787-Pm026572) million years ago (MYA). However, the largest divergence time of the duplicated UGTs between P. mume and P. persica was 72.16 MYA (Pm021307-ppa005517 m), followed by 56.04 MYA (Pm002464-ppa005187 m). Most duplicated gene pairs diverged around 1 to 5 MYA.

Phylogenetic Analysis of P. mume
These 130 putative PmUGTs and 112 A. thaliana UGTs (AtUGTs) were used for phylogenetic analysis to highlight the gene loss and gene gain events. Besides, 2 maize UGTs (GRMZM2G075387 and GRMZM5G834303) and 4 peach UGTs (Prupe.7G055200, Prupe.6G265900, Prupe.6G267000 and Prupe.6G266600) represented O and P groups were also added to identify PmUGTs O and P candidates. Phylogenetic result revealed that 14 groups, A to N, were clustered and no member was All the segmental duplicated UGT gene pairs had undergone a whole-genome duplication and the Ka/Ks ratios were less than 1. This result indicated that these UGTs experienced negative selection during species evolution process. Moreover, the divergence times of the duplicated UGTs at P. mume were significantly larger than that between P. mume and P. persica. The divergence time of the eight duplicated pairs at P. mume genome spanned from 47.84 (Pm025006-Pm030552) to 99.11 (Pm005787-Pm026572) million years ago (MYA). However, the largest divergence time of the duplicated UGTs between P. mume and P. persica was 72.16 MYA (Pm021307-ppa005517 m), followed by 56.04 MYA (Pm002464-ppa005187 m). Most duplicated gene pairs diverged around 1 to 5 MYA.

Phylogenetic Analysis of P. mume
These 130 putative PmUGTs and 112 A. thaliana UGTs (AtUGTs) were used for phylogenetic analysis to highlight the gene loss and gene gain events. Besides, 2 maize UGTs (GRMZM2G075387 and GRMZM5G834303) and 4 peach UGTs (Prupe.7G055200, Prupe.6G265900, Prupe.6G267000 and Prupe.6G266600) represented O and P groups were also added to identify PmUGTs O and P candidates.
Phylogenetic result revealed that 14 groups, A to N, were clustered and no member was identified in group O and P ( Figure 3). In each group, most of the UGT members were the same between P. mume and A. thaliana except in G (18 in P. mume and 6 in A. thaliana) and H (10 in P. mume and 19 in A. thaliana).  Prupe.6G267000 and Prupe.6G266600) were included. The full-length sequences of the UTG proteins were aligned using CLUSTALW and the phylogenetic tree was constructed using the ML method in the MEGA 6.0 [37]. The colored lines mark the groups of the UGTs.

Genomic Characteristics of the Putative UGTs in P. mume
We analyzed the exon/intron and conserved motif characteristics of the 130 PmUGTs to investigate their structural diversity. Among them, 70 UGTs possess at least 1 intron and 60 possess no introns. Of the 70 intron-containing UGTs, most UGTs had 1-4 introns, with a ratio of 1.44 introns per intron-containing UGTs. And Pm022854 contained the maximum number of introns (14), followed by Pm000211 with 7 introns. In each phylogenetic group, the intron numbers were different. The maximum number of introns was found in E, D, L and H, whereas the minimum number of introns was found in B, C, F, I, M and N groups. It is interesting that members within each group thaliana UGTs, 2 maize UGTs (GRMZM2G075387 and GRMZM5G834303) and 4 peach UGTs (Prupe.7G055200, Prupe.6G265900, Prupe.6G267000 and Prupe.6G266600) were included. The full-length sequences of the UTG proteins were aligned using CLUSTALW and the phylogenetic tree was constructed using the ML method in the MEGA 6.0 [37]. The colored lines mark the groups of the UGTs.

Genomic Characteristics of the Putative UGTs in P. mume
We analyzed the exon/intron and conserved motif characteristics of the 130 PmUGTs to investigate their structural diversity. Among them, 70 UGTs possess at least 1 intron and 60 possess no introns. Of the 70 intron-containing UGTs, most UGTs had 1-4 introns, with a ratio of 1.44 introns per intron-containing UGTs. And Pm022854 contained the maximum number of introns (14), followed by Pm000211 with 7 introns. In each phylogenetic group, the intron numbers were different. The maximum number of introns was found in E, D, L and H, whereas the minimum number of introns was found in B, C, F, I, M and N groups. It is interesting that members within each group exhibited similarity intron/extron genomic characteristics (Supplementary Figure S1). The same result was also obtained in conserved motifs structure ( Figure 4). These results suggested that PmUGT family members within group were relatively conserved and diverged greatly among different groups.  Figure S1). The same result was also obtained in conserved motifs structure ( Figure 4). These results suggested that PmUGT family members within group were relatively conserved and diverged greatly among different groups.

Transcriptome Analysis of Tissue-Specific Expression of PmUGTs
To detect the expression differences of PmUGTs, we analyzed their transcript abundances in bud, fruit, leaf, root and stem according to RNA-seq data. After filtering the low and missing expression values, 123 PmUGTs were finally examined to be expressed across the different tissues. Through hierarchical clustering analysis, these 123 PmUGTs were grouped into five discrete clusters in the five tested tissues ( Figure 5). The expressed 16 PmUGTs in cluster A showed consistent

Transcriptome Analysis of Tissue-Specific Expression of PmUGTs
To detect the expression differences of PmUGTs, we analyzed their transcript abundances in bud, fruit, leaf, root and stem according to RNA-seq data. After filtering the low and missing expression values, 123 PmUGTs were finally examined to be expressed across the different tissues. Through hierarchical clustering analysis, these 123 PmUGTs were grouped into five discrete clusters in the five tested tissues ( Figure 5). The expressed 16 PmUGTs in cluster A showed consistent downregulated expression patterns in bud, fruit, leaf and root but upregulated patterns in stem. The expression levels of 24 PmUGTs in cluster B were relatively higher in bud and fruit when compared with other tissues. In cluster C, 18 UGTs were detected to display upregulated expression level only in bud, with relatively low levels in other four tissues. PmUGTs in Cluster D, with the largest number of 37, exhibited upregulated expression in fruit; while 28 PmUGTs in Cluster E displayed upregulated expression in root and Cluster F showed high level in leaf ( Figure 5).

Transcriptome Analysis of PmUGTs Expression during Bud Dormancy Transition
In the present paper, we also displayed the PmUGTs expression profiles at four dormancy stages: EDI (with no flush sign in the phytotron), EDII (with 45% flush rate), EDIII (with 95% flush rate) and NF (natural flush). More details can be seen in Zhang et al. [37]. The expression profiles of PmUGTs during bud dormancy transition were hierarchically clustered into five groups ( Figure 6). Cluster A (including 19 PmUGTs) exhibited highest levels at EDII and then gradually decreased as dormancy release progressed. Nine genes in Cluster B had highest level at EDI stage and then maintained

Transcriptome Analysis of PmUGTs Expression during Bud Dormancy Transition
In the present paper, we also displayed the PmUGTs expression profiles at four dormancy stages: EDI (with no flush sign in the phytotron), EDII (with 45% flush rate), EDIII (with 95% flush rate) and NF (natural flush). More details can be seen in Zhang et al. [37]. The expression profiles of PmUGTs during bud dormancy transition were hierarchically clustered into five groups ( Figure 6). Cluster A (including 19 PmUGTs) exhibited highest levels at EDII and then gradually decreased as dormancy release progressed. Nine genes in Cluster B had highest level at EDI stage and then maintained relatively lower level at EDII, EDIII and NF stages. Cluster C genes showed highest expression level at EDI and sharply decreased at EDII and EDIII. After dormancy released, these genes subsequently increased at NF stage. The 10 PmUGTs in cluster D displayed relatively low expression levels at the EDI and EDII stages and then increased sharply at EDIII and maintained relatively high level at NF stage. Cluster E contained the large number of PmUGTs (80 genes). These genes showed relatively low expression level at dormancy stages and sharply increased once the dormancy completely released. relatively lower level at EDII, EDIII and NF stages. Cluster C genes showed highest expression level at EDI and sharply decreased at EDII and EDIII. After dormancy released, these genes subsequently increased at NF stage. The 10 PmUGTs in cluster D displayed relatively low expression levels at the EDI and EDII stages and then increased sharply at EDIII and maintained relatively high level at NF stage. Cluster E contained the large number of PmUGTs (80 genes). These genes showed relatively low expression level at dormancy stages and sharply increased once the dormancy completely released.

Quantitative Real-Time PCR Analyses of PmUGTs in Response to ABA Treatment
Previous studies revealed that UGTs from group E might participate in response to ABA stress. In this paper, 12 group E PmUGTs were identified and then RT-qPCR was employed to investigate the expression patterns under ABA treatment. According to the RNA-seq data, PmUGT61/Pm014846, PmUGT62/Pm014847 and PmUGT63/Pm014848 showed extremely low expression levels in bud, leaf, stem and fruit. Moreover, these three PmUGTs displayed low expression pattern and showed no

Quantitative Real-Time PCR Analyses of PmUGTs in Response to ABA Treatment
Previous studies revealed that UGTs from group E might participate in response to ABA stress. In this paper, 12 group E PmUGTs were identified and then RT-qPCR was employed to investigate the expression patterns under ABA treatment. According to the RNA-seq data, PmUGT61/Pm014846, PmUGT62/Pm014847 and PmUGT63/Pm014848 showed extremely low expression levels in bud, leaf, stem and fruit. Moreover, these three PmUGTs displayed low expression pattern and showed no change during bud dormancy process. Thereafter, in this study, we only investigated the expression patterns of the other nine PmUGTs under ABA treatment in leaves. Of the nine PmUGT genes, seven genes were obviously up-regulated in response to ABA stress, while the remaining two genes PmUGT5/Pm000549 and PmUGT46/Pm010818 showed slight expression changes (<2-fold) (Figure 7). It is interesting that all seven selected PmUGTs were up-regulated at early stages and then down regulated after reaching a peak expression. The PmUGT56/Pm014838, PmUGT57/Pm014839, PmUGT59/Pm014843 and PmUGT60/Pm014844 were strongly up-regulated at 4 h after ABA treatment by more than 10-fold compared to the control, whereas, their expression revealed relatively decreased thereafter. Besides PmUGT58/Pm014842, PmUGT54/Pm014833 and PmUGT55/Pm014836 were slightly up-regulated by about 5-fold compared to the control. It is noting that these three PmUGTs showed different peaking time, suggesting their roles might be slightly different (Figure 7). change during bud dormancy process. Thereafter, in this study, we only investigated the expression patterns of the other nine PmUGTs under ABA treatment in leaves. Of the nine PmUGT genes, seven genes were obviously up-regulated in response to ABA stress, while the remaining two genes PmUGT5/Pm000549 and PmUGT46/Pm010818 showed slight expression changes (<2-fold) (Figure 7). It is interesting that all seven selected PmUGTs were up-regulated at early stages and then down regulated after reaching a peak expression. The PmUGT56/Pm014838, PmUGT57/Pm014839, PmUGT59/Pm014843 and PmUGT60/Pm014844 were strongly up-regulated at 4 h after ABA treatment by more than 10-fold compared to the control, whereas, their expression revealed relatively decreased thereafter. Besides PmUGT58/Pm014842, PmUGT54/Pm014833 and PmUGT55/Pm014836 were slightly up-regulated by about 5-fold compared to the control. It is noting that these three PmUGTs showed different peaking time, suggesting their roles might be slightly different (Figure 7).

Discussion
Increasing studies indicate that UGT proteins play important roles in plant growth and adaptation to environmental stress. In addition, UGT is involved in carbohydrate metabolism during the bud dormancy release process [36]. To our knowledge, no further information is available about the UGT gene family in P. mume. Here, we conducted a comprehensive investigation of the PmUGT family.
Plant genomes contain numerous UGT genes, with UGT members varying among species. For

Discussion
Increasing studies indicate that UGT proteins play important roles in plant growth and adaptation to environmental stress. In addition, UGT is involved in carbohydrate metabolism during the bud dormancy release process [36]. To our knowledge, no further information is available about the UGT gene family in P. mume. Here, we conducted a comprehensive investigation of the PmUGT family.
Phylogenetic analysis consistently clustered 14 distinct groups (A-N) with Arabidopsis [9]. This result indicates that the UGT family in P. mume has not phylogenetically diversified after separation from Arabidopsis. In some species of Rosaceae, including peach, apple and grape, 16 distinct phylogenetic groups (A-P) are known, while 17 groups (A-Q) have been observed in Z. mays [10,12,13,38]. The O and P groups found in peach and maize are absent in P. mume, which suggests they were lost at some stage during evolution. Surprisingly, PmUGT members in different groups, except for those in G and H, were similar to members of corresponding groups in Arabidopsis, which suggests that they have a conserved substrate specificity. Groups E, G and H were reduced in P. mume relative to peach, which indicates that these UGTs may be less critical in P. mume.
PmUGTs exhibit tissue-specific expression patterns. Determining whether the expressed UGT genes are functionally diverged or conserved should improve our understanding of plant adaptation to changing environments [4]. PmUGT2, PmUGT42, PmUGT77, PmUGT80, PmUGT98, PmUGT105, PmUGT120 and PmUGT121 were expressed at relatively high levels in all tested tissues, suggesting their involvement in overall tissue development process. PmUGT17, PmUGT28, PmUGT36, PmUGT43, PmUGT49, PmUGT50, PmUGT51, PmUGT53, PmUGT73, PmUGT74 and PmUGT122 were expressed at extremely low or undetectable levels in all tissues, which suggests that these genes do not play an important role in P. mume development. PmUGT120 and PmUGT32 was specifically highly expressed in leaf and in root, which implies that these genes may have a specific function in leaf and root, respectively. The same result was observed in peach. Prupe.1G091100 and Prupe.1G091000 (homologs of Pm027780 and Pm019616, respectively) were mainly expressed in peach flowers [41]. These two UGTs are responsible for anthocyanin synthesis in peach flowers [41]. The dynamic expression patterns of several hormone-related UGTs, such as Pm014836 (UGT71B6, associated with ABA), Pm030035 (UGT74B1, IAA), Pm014886 (UGT85A1, CK) and Pm026307 (UGT73C1, CK), suggest that hormone conjugation plays important roles during the P. mume dormancy process. Even closely related homologs exhibited different spatial-and tissue-specific expression patterns. For example, the AtUGTs, UGT71B6, UGT71B7 and UGT71B8 exhibited very high expression levels in leaves, flowers and siliques, respectively [42].
Phytohormones play crucial roles in the regulation of protective responses against biotic and abiotic stresses but the mechanism of hormone glycosylation remains poorly understood. The availability of data from Arabidopsis provides sufficient information about the UGT family and several UGT genes have been functionally characterized as the glycoconjugates of phytohormones. For example, the AtUGTs, UGT75D1, UGT71C5 and UGT71B6 glycosylate ABA; UGT74B1, UGT74D1 and UGT84B1, glycosylate IAA. In addition, UGT74F1, UGT73B3 and UGT73B5 participate in SA glycosylation, while UGT76B1 is involved in crosstalk between SA and JA [21,31,[43][44][45].
ABA is an important hormone regulating plant development and adaptive responses but information regarding ABA homeostasis is limited. The fine-tuning of ABA biosynthetic and catabolic pathways is crucial for balancing cellular ABA levels [1]. Cellular ABA content is lowered via two pathways, hydroxylation and conjugation [44][45][46][47][48]. In the first pathway, cytochrome P450 monooxygenase hydroxylates ABA at the C-80 position to form unstable 80-hydroxy ABA that is converted to phaseic acid. In the second pathway, ABA and hydroxy ABA are conjugated with glucose for inactivation [23,47,48]. It is the ABA glucosyltransferase that performs the conjugation and ABA-GE is the predominant form [23]. It is reported that ABA-GE can be transported between tissues and in some tissues, conjugation is the major pathway of ABA inactivation. Meanwhile, ABA-GE provided an ABA source for subsequent hydrolysis [43,49,50].
Several ABA-related UGTs and their close homologs have been functionally characterized, which can inactivate ABA and lower ABA levels. For example, ABA glycosylation by UGT71B6, UGT71B7 and UGT87A2 has been well documented in Arabidopsis, with this function also reported for UGT71A33 and UGT71A35 in strawberry and ABAGT in Vigna angularis [21,22,43,44]. As inferred by the suppression of RD29Ap:LUC, UGT71B6, UGT71B7 and UGT71B8 reduce cellular ABA levels. UGT RNAi (triple knock-out mutant) transgenic plants are sensitive to exogenous ABA and salinity stress during seed germination and subsequent development process. In contrast, the over-expression of UGT71B6 in an atbg1 mutant background aggravates the ABA-deficient phenotype [42]. In the present study, 12 UGT71B6 homologs were identified and placed in group E. We examined the transcript levels of nine of these PmUGTs under exogenous ABA treatment. As shown in Figure 7, seven PmUGTs were significantly upregulated by exogenous ABA treatment, albeit at different levels. The other two PmUGTs were only slightly changed. This result indicates that these UGTs are involved in ABA glucosylation in P. mume.

Genome-Wide Identification of UGT Family Genes in P. mume
To identify the candidate UGT genes in P. mume, a total of 120 Arabidopsis UGT protein sequences were retrieved from CAZy (available online: http://www.cazy.org/GlycosylTransferases.html) and 168 peach UGT proteins were downloaded from Phytozome V12.1 (available online: https: //phytozome.jgi.doe.gov/pz/portal.html). All these sequences were used as query to BLASTP against P. mume proteome with a cut-off E-value of 1 × 10 −10 . Subsequently, the conserved PSPG box sequence was also used as a query to BLASTP against P. mume proteome database. Furthermore, the Hidden Markov Model (HMM) profile of UDPGT domain (PF00201) was retrieved from Pfam 29.0 (available online: http://pfam.xfam.org/) and used to search against the P. mume proteome database. The amino acid sequences of candidates from these three strategies were screened by SMART (available online: http://smart.emblheidelberg.de) to remove proteins without a complete PSPG box.

PmUGT Genes Location and Characteristics
InterPro was used to check the validation of final UGT genes [51]. The ORF and chromosome distribution of P. mume UGTs was obtained from P. mume genome database. MapChart (v2.3) was used to visualize the chromosomal location of PmUGTs [52]. ExPASy (available online: http://expasy.org/) was used to estimate the isoelectric point and molecular weight. The subcellular localization of each PmUGT was analyzed using the CELLO v2.5 server (available online: http://cello.life.nctu.edu.tw/).

Analyses of Gene Structure and Conserved Motifs of UGT Genes
According to the general feature format file of P. mume, the exon-intron structures of the PmUGTs were obtained and graphed with the Gene Structure Display Server (GSDS: available online: http: //gsds.cbi.pku.edu.ch). The conserved motifs of the putative UGT proteins were predicted by using the on-line MEME procedure with maximum 15 motifs per sequence. The sequence logo was obtained using the online Weblogo platform (available online: http://weblogo.berkeley.edu).

Homology Analysis and Selection Pressures of UGT Gene Pairs between P. mume and P. persica
To estimate the divergence of the putative tandem-duplicated UGT genes between P. mume and P. persica, the duplicated pairs were detected in the Plant gene duplication database (available online: http://chibba.agtec.uga.edu/duplication/). Mcscan [53] was employed to identify homologous regions and syntenic blocks were evaluated using Circos-0.64 [54]. The ratios of Ka (non-synonymous)/Ks (synonymous rate) of UGT gene pairs between P. mume and P. persica were calculated to estimate selection modes by using PAML software. 1.5 × 10 −8 was taken as synonymous substitutions per site per year in the case of dicotyledonous plants for MYA calculation. The Ka/Ks ratios greater than 1, equal to 1 and less than 1 represent positive, neutral and negative selection, respectively.

Transcriptome Analysis for Tissue-Specific Expression
To check tissue-specific expression of the putative UGTs in P. mume, the RNA-Seq data in different tissues, such as flower, leaves, roots and stem, were obtained. Besides, the transcript data at four crucial dormancy stages were also retrieved as detailed described by Zhang et al. [36]. The expression values for each PmUGT were calculated by fragments per kilobase of the exon model per million mapped reads by using the RNA-seq data of P. mume. The heat-maps of PmUGTs were established using R packages "heatmap".

RT-qPCR Analyses of the PmUGTs in Response to ABA Treatment
The seeds of P. mume were collected on cultivar "Lve" grown in the Jiufeng International Plum Blossom Garden, Beijing, China (40 • 07 N, 116 • 11 E). The seeds were sterilized with 20% sodium hypochlorite, washed with sterile water three times and were stored in the sand under 4 • C to promote germination. After three months, germinated seedlings were transplanted in nutritional soil in the greenhouse. For hormone treatment, 100 µM ABA were sprayed on the young seedlings until dropped. Fresh leaves were collected at 0, 1, 2, 4, 8, 12, 24 and 48 h, respectively. Samples were frozen in liquid nitrogen and then stored at −80 • C until used. Total RNA extraction and qPCR were performed as described in Zhang et al. [37] Primers sequences were listed in Supplementary Table S3.

Conclusions
A total of 130 PmUGTs were identified and clustered into 14 groups based on phylogenetic analysis and their chromosomal locations, gene structure, duplication events and conserved motifs were further investigated. RNA-seq analysis revealed specific expression patterns in different tissues. In addition, various changes in transcript levels were detected during bud dormancy release. We also uncovered differential responses of PmUGT expressions to ABA treatment using RT-qPCR. A major future research challenge is obtaining a better understanding of how plants regulate UGT members during development and in response to abiotic and biotic stress. Exploring the crosstalk between UGTs and other genes/proteins is also necessary. Our results provide important information on the UGT family in P. mume that will aid the further characterization of their biological roles in response to environmental stress.