Identification and Expression Analysis of Polygalacturonase Family Members during Peach Fruit Softening

Polygalacturonase (PG) is an important hydrolytic enzyme involved in pectin degradation during fruit softening. However, the roles of PG family members in fruit softening remain unclear. We identified 45 PpPG genes in the peach genome which are clustered into six subclasses. PpPGs consist of four to nine exons and three to eight introns, and the exon/intron structure is basically conserved in all but subclass E. Only 16 PpPG genes were expressed in ripening fruit, and their expression profiles were analyzed during storage in two peach cultivars with different softening characteristics. Eight PGs (PpPG1, -10, -12, -13, -15, -23, -21, and -22) in fast-softening “Qian Jian Bai” (QJB) fruit and three PGs (PpPG15, -21, and -22) in slow-softening “Qin Wang” (QW) fruit exhibited softening-associated patterns; which also were affected by ethylene treatment. Our results suggest that the different softening characters in QW and QJB fruit is related to the amount of PG members. While keeping relatively lower levels during QW fruit softening, the expression of six PGs (PpPG1, -10, -12, -11, -14, and -35) rapidly induced by ethylene. PpPG24, -25 and -38 may not be involved in softening of peach fruit.


Introduction
Fruit maturation is a complex and highly coordinated developmental process. In fleshy fruits, the main changes associated with ripening affect color, firmness, taste and flavor, and increase not only palatability, but also susceptibility to physical damage and shortening of storage life [1,2]. Peach is a typical climacteric fruit, and rapid softening and short shelf life after harvest adversely affect its market value [3]. Therefore, studying the physiological and molecular processes governing the softening of peach would be greatly beneficial in extending its shelf life.
Modification of the cell wall is thought to underlie the changes in fruit firmness and texture [4]. Plant cell wall is a complex reticulate structure, being composed of cellulose, hemicellulose, pectin and structural proteins [5,6]. Among these four macromolecular substances, pectin is the major component in the middle lamella and cell primary wall and can link cells together like "glue" [7][8][9][10]. Thus, systematic disassembly of pectin is critical to some developmental processes in plants, such as organ abscission, fruit ripening, and pod and anther dehiscence [11,12]. The degradation of pectin is catalyzed by hydrolytic enzymes, mainly including polygalacturonase (PG), pectate lyase, pectin methylesterases and β-galactosidase, of which PG has been suggested to play a central role [11,13].
Involvement of PG in fruit softening has been widely reported. In strawberry, silencing of FaPG1 gene decreases the breakdown of the middle lamella and slows the fruit softening [14]. In pear and banana, fruits softening are positively regulated by Pc-PG1 and Pc-PG2, and MaPG3 and MaPG4, respectively, with a pattern of ethylene-dependent [15,16]. The similar results have also been observed in papaya, guava, capsicum and apple [17][18][19][20]. However, Smith et al. thought PG was not essential for tomato fruit softening [21]. Interestingly, peach fruit melting and freestone are also controlled by Endo-PGs [22][23][24][25]. In addition, PG genes also participate in the organ abscission of lychee, oilseed rape, Arabidopsis and tomato [26][27][28]. In Arabidopsis, QUARTET2 (QRT2) is involved in pollen grain separation [29], and ADPG1 and ADPG2 are related to silique dehiscence and floral organ abscission [30].
PGs belong to one of the largest hydrolase families in plants [31], and the PG gene family has been identified in various plant species, including Arabidopsis [29], Oryza sativa [31], Populus [32], Cucumis sativus and Citrullus lanatus [33]. The expression of 66 PG genes of Arabidopsis has been tested by RT-PCR in five tissues, flowers, siliques, stems, leaves and roots, but only 43 had PCR products, 40 in flower, 34 in silique and root, 30 in leaf and 31 in stem [31]. In Populus, 11 PG members are related to flower development and two are related to leaf abscission under salt stress [32]. In Cucumis sativus, most CsPGs exhibit specific or high expression levels in a particular organ or tissue [33]. These studies have demonstrated that PG members are differentially expressed in specific tissues and in response to specific treatments, indicating extensive functional divergence among plant PG genes. Although the important roles of PG hydrolase in peach softening were confirmed by many researches [3,4,34], only a small number of PG family members have been reported [25,[34][35][36]. To clarify which PG members are involved in softening of peach fruit, in this study, we identify 45 members of the PG family from the whole genomic sequence of peach and analyze the expression profiles of PpPG genes during fruit softening in two peach cultivars that have different softening characteristics. We identify a small subset of PG genes that display consistent softening-associated expression patterns. We investigate the connection between selected ripening-and softening-related PG genes and the mechanism underlying ethylene-dependent ripening.

Identification of Polygalacturonase (PG) Family Members in Peach
The Hidden Markov Model (HMM) profile of the glycosyl hydrolase family 28 (GH28) domain was used as a query to blast against the NCBI non-redundant protein database and the Peach Genome Database. All reported Arabidopsis PG proteins were also used to search against the peach genome database using NCBI tblastn. Then, all candidate PG family members in the peach genome database were identified using four highly conserved PG domains. Forty-five PG genes were identified in peach and named PpPG1-45 according to their chromosomal location (Table 1) (Table 1).

Phylogenetic Analysis of PG Family Members in Peach
The phylogenetic tree was created using MEGA 6.0 with full-length PG protein sequences from peach, Arabidopsis thaliana and some selected fruit species. The 45 peach PG proteins were clustered into six subclasses (subclasses A to F). Subclass G was Arabidopsis QRT3PG, which is involved in degradation of the pollen mother cell wall [37], and did not include any peach PG members. Most peach PG members were included in subclasses C and E, while fewer members were included in subclasses B and D. PpPG21 and -22 were similar to PdPG; PpPG12 and -15 were similar to LcPG; and PpPG1 and -10 were similar to VvPG1 and CmPG1, respectively ( Figure 2). We defined the most likely orthologs between peach and Arabidopsis. This suggested that there were at least 25 ancestral PGs before the divergence of other species and peach ( Figure 2, red circles). Three of these nodes in subclasses of B, E and F, respectively ( Figure 2, black circles), however, had relatively low bootstrap support (<50%).

Genome Distribution and Gene Structures of PG Family Members in Peach
PG gene family members are distributed on all peach chromosomes. Chr 7 contains 14 PpPG genes; chr 1 and chr 4, and chr 6 and chr 8 contain seven PpPG genes and three PpPG genes, respectively; chr 2 contains four; chr 3 contains five; and chr 5 contains two. We can also observe tandem duplication sites such as PpPG8 and -7 on chr 1, and PpPG19, -20, -21 and -22 on chr 4 ( Figure 3).
All peach PG protein sequences were used to construct an unrooted phylogenetic tree with MEGA 6.0, and it was consistent with the rooted tree described above ( Figure 4). The exon/intron structures and intron phase of peach PG genes were analyzed using the online tool GSDS, using their full-length CDS sequences and corresponding genomic DNA sequences. The results showed that PpPGs consisted of four to nine exons and three to eight introns, and the members of subclasses A, B and F generally had more exons and introns. In addition, the exon/intron structure is basically conserved in each subclass except for subclass E (Figure 4).
To further investigate conserved motifs in the amino acid sequences of peach PG genes, 45 peach PG proteins were aligned using the online tool MEME set to output eight motifs. Motif 1 was found in all 45 peach PG members, while motif 4 was absent in PpPG32, motif 3 was absent in PpPG9 and -37, and motif 7 was absent in PpPG32 and -45. Some motifs only existed in certain members, such as motif 2, which existed in most PpPG members, but not in subclass E. Moreover, most members of subclass E were also missing motif 5 and motif 8. Furthermore, PpPG44 in subclass B did not have motifs 2, 5 or 8; PpPG9 in subclass C did not have motifs 3 or 6, and PpPG32 had only four motifs ( Figure 4).

Genome Distribution and Gene Structures of PG Family Members in Peach
PG gene family members are distributed on all peach chromosomes. Chr 7 contains 14 PpPG genes; chr 1 and chr 4, and chr 6 and chr 8 contain seven PpPG genes and three PpPG genes, respectively; chr 2 contains four; chr 3 contains five; and chr 5 contains two. We can also observe tandem duplication sites such as PpPG8 and -7 on chr 1, and PpPG19, -20, -21 and -22 on chr 4 ( Figure 3).
All peach PG protein sequences were used to construct an unrooted phylogenetic tree with MEGA 6.0, and it was consistent with the rooted tree described above ( Figure 4). The exon/intron structures and intron phase of peach PG genes were analyzed using the online tool GSDS, using their full-length CDS sequences and corresponding genomic DNA sequences. The results showed that PpPGs consisted of four to nine exons and three to eight introns, and the members of subclasses A, B and F generally had more exons and introns. In addition, the exon/intron structure is basically conserved in each subclass except for subclass E (Figure 4).
To further investigate conserved motifs in the amino acid sequences of peach PG genes, 45 peach PG proteins were aligned using the online tool MEME set to output eight motifs. Motif 1 was found in all 45 peach PG members, while motif 4 was absent in PpPG32, motif 3 was absent in PpPG9 and -37, and motif 7 was absent in PpPG32 and -45. Some motifs only existed in certain members, such as motif 2, which existed in most PpPG members, but not in subclass E. Moreover, most members of subclass E were also missing motif 5 and motif 8. Furthermore, PpPG44 in subclass B did not have motifs 2, 5 or 8; PpPG9 in subclass C did not have motifs 3 or 6, and PpPG32 had only four motifs ( Figure 4). insititia (PdPG), Pyrus communis (PcPG), Pyrus bretschneideri (PbPG), Pyrus pyrifolia (PpyPG), Solanum lycopersicum (SlPG), Elaeis guineensis (EgPG) and Vitis vinifera (VvPG). Colored triangles and squares represent PpPGs and each color indicates a subclass; triangles indicate PpPG genes that are expressed in peach fruit and squares represent PpPGs that are not expressed in peach fruit according to results of RNA-seq. Red and black nodes indicate that the bootstrap support for the branches defining the common ancestral PGs is more than 50% and less than 50%, respectively. Bootstrap values lower than 50 are hidden in the unrooted phylogenetic tree. Solanum lycopersicum (SlPG), Elaeis guineensis (EgPG) and Vitis vinifera (VvPG). Colored triangles and squares represent PpPGs and each color indicates a subclass; triangles indicate PpPG genes that are expressed in peach fruit and squares represent PpPGs that are not expressed in peach fruit according to results of RNA-seq. Red and black nodes indicate that the bootstrap support for the branches defining the common ancestral PGs is more than 50% and less than 50%, respectively. Bootstrap values lower than 50 are hidden in the unrooted phylogenetic tree.

Fruit Firmness, Ethylene Production and PG Activity Change during Softening
The two cultivars have different fruit softening characteristics: QJB fruit softening rapidly during ripening, and QW fruits remain firm for a long time during ripening. The firmness, ethylene production, and PG activity in QJB and QW fruits were measured during fruit storage. For QJB fruit, flesh firmness decreased slowly in the first two days after harvest (DAH), and decreased rapidly from two to four DAH, then declined slowly again ( Figure 5A). For QW fruit, flesh firmness remained stable for the first 20 DAH, and decreased slowly from 20 to 22 DAH, then declined dramatically from 22 to 24 DAH ( Figure 5B). Ethylene production was very low in QW fruit for the first 22 DAH and then increased markedly from 22 to 24 DAH ( Figure 5D). In QJB fruit, ethylene production increased slowly in the first 2 DAH, and increased rapidly from two to six DAH, then decreased considerably ( Figure 5C).

Fruit Firmness, Ethylene Production and PG Activity Change during Softening
The two cultivars have different fruit softening characteristics: QJB fruit softening rapidly during ripening, and QW fruits remain firm for a long time during ripening. The firmness, ethylene production, and PG activity in QJB and QW fruits were measured during fruit storage. For QJB fruit, flesh firmness decreased slowly in the first two days after harvest (DAH), and decreased rapidly from two to four DAH, then declined slowly again ( Figure 5A). For QW fruit, flesh firmness remained stable for the first 20 DAH, and decreased slowly from 20 to 22 DAH, then declined dramatically from 22 to 24 DAH ( Figure 5B). Ethylene production was very low in QW fruit for the first 22 DAH and then increased markedly from 22 to 24 DAH ( Figure 5D). In QJB fruit, ethylene production increased slowly in the first 2 DAH, and increased rapidly from two to six DAH, then decreased considerably ( Figure 5C). The activities of exo-PG and endo-PG in the two varieties showed similar trends during storage. In QJB fruit, the activities of exo-PG and endo-PG remained stable for the first two DAH, then increased and reached their maximum activity at 4 DAH, and declined from four to six DAH ( Figure 6A,C). In QW fruit, the activities of exo-PG and endo-PG also remained at a stable level in the first 20 DAH, then increased and reached their maximum activity at 24 DAH and 22 DAH, respectively, and then declined ( Figure 6B,D). Furthermore, the rate of change of endo-PG activity in QJB fruit was significantly higher than that in QW fruit during storage ( Figure 6). The activities of exo-PG and endo-PG in the two varieties showed similar trends during storage. In QJB fruit, the activities of exo-PG and endo-PG remained stable for the first two DAH, then increased and reached their maximum activity at 4 DAH, and declined from four to six DAH ( Figure 6A,C). In QW fruit, the activities of exo-PG and endo-PG also remained at a stable level in the first 20 DAH, then increased and reached their maximum activity at 24 DAH and 22 DAH, respectively, and then declined ( Figure 6B,D). Furthermore, the rate of change of endo-PG activity in QJB fruit was significantly higher than that in QW fruit during storage ( Figure 6).

Fruit Firmness, Ethylene Production and PG Activity Change during Softening
The two cultivars have different fruit softening characteristics: QJB fruit softening rapidly during ripening, and QW fruits remain firm for a long time during ripening. The firmness, ethylene production, and PG activity in QJB and QW fruits were measured during fruit storage. For QJB fruit, flesh firmness decreased slowly in the first two days after harvest (DAH), and decreased rapidly from two to four DAH, then declined slowly again ( Figure 5A). For QW fruit, flesh firmness remained stable for the first 20 DAH, and decreased slowly from 20 to 22 DAH, then declined dramatically from 22 to 24 DAH ( Figure 5B). Ethylene production was very low in QW fruit for the first 22 DAH and then increased markedly from 22 to 24 DAH ( Figure 5D). In QJB fruit, ethylene production increased slowly in the first 2 DAH, and increased rapidly from two to six DAH, then decreased considerably ( Figure 5C). The activities of exo-PG and endo-PG in the two varieties showed similar trends during storage. In QJB fruit, the activities of exo-PG and endo-PG remained stable for the first two DAH, then increased and reached their maximum activity at 4 DAH, and declined from four to six DAH ( Figure 6A,C). In QW fruit, the activities of exo-PG and endo-PG also remained at a stable level in the first 20 DAH, then increased and reached their maximum activity at 24 DAH and 22 DAH, respectively, and then declined ( Figure 6B,D). Furthermore, the rate of change of endo-PG activity in QJB fruit was significantly higher than that in QW fruit during storage ( Figure 6).

Identification of PG Genes Exhibiting Ripening-Associated Patterns of Expression
We identified the PG genes expressed in fruit by transcriptome sequencing of QJB fruit at the ripening stage, and the results showed that among 45 PpPG genes, only 16 were expressed in peach fruit (Table S1) (Table S1).
The transcript accumulation patterns of the 16 PpPG genes expressed in peach fruit were assessed in the two varieties by qRT-PCR. The results were consistent with the transcriptome data, and nine PpPG genes (PpPG1, -10, -12, -13, -15, -21, -22, -23 and -25) showed up-regulation, but some differences existed between the two varieties. For QJB fruit, the expression of PpPG1, -10, -12, -21 and -22 was dramatically and continuously increased during the whole storage period, and reached its maximum at the end of storage. PpPG15, -23 and -13 displayed up-regulation for the first four DAH, and then maintained a stable level or declined rapidly. PpPG25 expression remained stable in early storage, and then increased continuously (Figure 7).
Compared with QJB, the expression of PpPG1, -10, -12 and -15 in QW fruit showed a lower level of transcript abundance during the whole storage period, with elevated expression levels only from 22 to 24 DAH. PpPG13 expression was undetectable in QW during the whole storage period. Expression of PpPG21 and -22 remained at a very low level for the first 20 DAH, coinciding with the stable firmness. The expression levels subsequently dramatically increased from 20 DAH to 24 or 28 DAH. The expression of PpPG23 was maintained stably at a high level during the first 22 DAH, increased markedly from 22 to 24 DAH, and then decreased rapidly. PpPG25 expression remained relatively stable during storage except at 16 DAH (Figure 7).
During storage, the expression of PpPG11 and -14 exhibited declining trends in the two varieties ( Figure 7). PpPG24, -35 and -38 also showed similarity in expression abundance and patterns in the two varieties during storage, and expression of these genes remained relatively stable (Figure 7). PpPG6 and -39 expression was undetectable in the two varieties.

Identification of PG Genes Exhibiting Ripening-Associated Patterns of Expression
We identified the PG genes expressed in fruit by transcriptome sequencing of QJB fruit at the ripening stage, and the results showed that among 45 PpPG genes, only 16 were expressed in peach fruit (Table S1) (Table S1).
The transcript accumulation patterns of the 16 PpPG genes expressed in peach fruit were assessed in the two varieties by qRT-PCR. The results were consistent with the transcriptome data, and nine PpPG genes (PpPG1, -10, -12, -13, -15, -21, -22, -23 and -25) showed up-regulation, but some differences existed between the two varieties. For QJB fruit, the expression of PpPG1, -10, -12, -21 and -22 was dramatically and continuously increased during the whole storage period, and reached its maximum at the end of storage. PpPG15, -23 and -13 displayed up-regulation for the first four DAH, and then maintained a stable level or declined rapidly. PpPG25 expression remained stable in early storage, and then increased continuously (Figure 7).
Compared with QJB, the expression of PpPG1, -10, -12 and -15 in QW fruit showed a lower level of transcript abundance during the whole storage period, with elevated expression levels only from 22 to 24 DAH. PpPG13 expression was undetectable in QW during the whole storage period. Expression of PpPG21 and -22 remained at a very low level for the first 20 DAH, coinciding with the stable firmness. The expression levels subsequently dramatically increased from 20 DAH to 24 or 28 DAH. The expression of PpPG23 was maintained stably at a high level during the first 22 DAH, increased markedly from 22 to 24 DAH, and then decreased rapidly. PpPG25 expression remained relatively stable during storage except at 16 DAH (Figure 7).
During storage, the expression of PpPG11 and -14 exhibited declining trends in the two varieties ( Figure 7). PpPG24, -35 and -38 also showed similarity in expression abundance and patterns in the two varieties during storage, and expression of these genes remained relatively stable (Figure 7). PpPG6 and -39 expression was undetectable in the two varieties.

Ethylene and 1-MCP Alter PG Gene Expression
To shed more light on the potential role of PGs in fruit ripening, we compared their expression under ethylene and/or 1-MCP treatment during storage. After ethylene treatment, the changes in firmness and ethylene release in the two varieties were very similar. The decrease in fruit firmness was faster than that in the control fruit, especially in the first two DAH. At the same time, after ethylene treatment, the rate of ethylene release increased sharply and peaked at two DAH, with higher peak rates for treated fruit than for controls ( Figure 5). Furthermore, in QJB fruit, after ethylene treatment, the expression of eight PpPG genes (PpPG1, -12, -13, -15, -21, -22, -24 and -25) also dramatically increased and peaked at two DAH, with higher peak expression for treated fruit than for controls (Figure 7). The expression of PpPG10 and -23 was also significantly increased by ethylene treatment, and peaked at four DAH. Compared with the control, ethylene treatment had

Ethylene and 1-MCP Alter PG Gene Expression
To shed more light on the potential role of PGs in fruit ripening, we compared their expression under ethylene and/or 1-MCP treatment during storage. After ethylene treatment, the changes in firmness and ethylene release in the two varieties were very similar. The decrease in fruit firmness was faster than that in the control fruit, especially in the first two DAH. At the same time, after ethylene treatment, the rate of ethylene release increased sharply and peaked at two DAH, with higher peak rates for treated fruit than for controls ( Figure 5). Furthermore, in QJB fruit, after ethylene treatment, the expression of eight PpPG genes (PpPG1, -12, -13, -15, -21, -22, -24 and -25) also dramatically increased and peaked at two DAH, with higher peak expression for treated fruit than for controls (Figure 7). The expression of PpPG10 and -23 was also significantly increased by ethylene treatment, and peaked at four DAH. Compared with the control, ethylene treatment had little effect on the expression of PpPG11, -14, -35 and -38. For slow-ripening QW fruit, after ethylene treatment, the expression of seven PpPG genes (PpPG1, -10, -12, -14, -15, -21 and -22) increased dramatically and peaked at two DAH, with higher peak expression for treated fruit than for controls (Figure 7). Ethylene treatment also significantly enhanced the expression levels of PpPG11 and -35, and had little effect on the expression of PpPG13, -23, -24, -25, and -38 (Figure 7).

PG Family Member Identification and Sequence Analysis
PG is a cell wall hydrolytic enzyme with important roles in many physiological processes, such as fruit ripening, seed germination, cell elongation, organ abscission and pollen tube elongation [4,12,14,20,26,30,38]. PGs are encoded by a typical large gene family in plants, and the PG gene family has been identified and characterized in some plants such as Arabidopsis, Populus, rice, Cucumis sativus and Citrullus lanatus [29,[31][32][33]. In this study, we identified 45 PG genes in the peach genome, and Arabidopsis, Populus, Cucumis sativus and Citrullus lanatus have 68, 75, 53 and 62 PG genes, respectively. Peach has fewer PG members than the above species, and this may be because peach has not undergone recent whole-genome duplication [39] and the expansion of subclasses is different in different plants.
Most PG members from various species contain four conserved domains [33,40]. Domains I and II are likely to compose the catalytic site, domain III can take part in catalytic reactions, and domain IV constitutes a likely candidate for ionic interactions with carboxylate groups present in the substrate [41]. The peach PG genes of subclasses A, B, C, D and F contain the four conserved domains and seven to nine motifs, but PG members of subclass E lack domain III and contain five or six motifs (Table 1, Figure 4). Other studies in plants have also shown that domain III of PGs in subclass E is not conserved [33,40]. Our results showed that most peach PG genes of subclass E are expressed in the ripening fruit (Figure 2), which is similar to the results in Arabidopsis [31]. In addition, the involvement of PG genes of E subclass in fruit abscission was also observed in lychee and Elaeis guineensis [26,42] Therefore, most fruit related PG genes may belongs to E subclass, which may be related to their conserved functions in fruit abscission and/or ripening. Except for very low expression of PpPG39, peach PG genes of subclasses D and F are not expressed in ripening fruit, which suggests that these PG genes cannot be involved in peach fruit softening during ripening. In addition, PpPG7, -8, -18, -26, -34, -36 in subclass F formed a special sub-clades without other species members, and had the low bootstrap support (<50%), which suggesting these PG members may be peach-special. Subclass C contained 14 peach PG members; however, only PpPG21 and -22 were expressed in the ripening fruit. PpPG21 and -22 are 100% similar to PdPG1 (ABD33834), which involved in the regulation of seed maturation and drying, fruit ripening in damson plum [43]. In addition, tomato TAPG1, -2, -4, and -5 in subclass C were associated with fruit abscission [44,45]. Similarly, Monocot oil palm EgPG4 gene in subclass C participated in cell wall pectin modifications as well during both mesocarp ripening and fruit shedding [42]. These studies suggested that PpPG21 and -22 are probably related to peach fruit ripening and abscission. Two tomato fruit ripening related PG genes (AAA34178, and NP_001234021) were classified into Subclass B [46,47], together with other fruit ripening related PG genes (such as, VvPG1 in grape and ZMdPG1 in apple) [48,49]. Meanwhile, in Arabidopsis, ADPG1 and -2 in subclass B had been reported to be essential for silique dehiscence, and ADPG2 and QRT2 contribute to floral organ abscission [30]. Our results also showed that PpPG10, -14, and -35 in group B were expressed in mature fruit, suggesting these PGs may be associated with peach fruit ripening.

Possible Roles of PG Family Members in Fruit Softening during Ripening
The role of PG in fruit softening has been studied extensively, and an increase in PG activity and mRNA levels has been observed during ripening in several fruits [4,11,[15][16][17]20,50]. PG hydrolyzes pectin acid along with the its main chain, causing pectin degradation, cell wall dissolution, and ultimately, fruit softening [51]. However, as a multi-gene family, PG members were expressed at different stages and in different tissues during the development of plants [31]. Thus far, little is known about the role of PG family members in fruit softening. Pan et al. (2015) indicated that PpPG21 (ppa006839) and -22 (ppa006857) were relatively highly expressed in ripening peach fruit, but that their transcripts were almost undetectable in fruit during development [52]. Gu et al. (2016) also demonstrated that PpPG22 and -21 are responsible for stone adhesion and melting flesh, respectively [25]. The mRNA levels of PpPG1 (ppa018113) were lower in both the melting type and the stony hard type peach, although they increased slightly after ethylene treatment in the stony hard type peach [36].
Based on the expression profiles of 16 PpPG genes during fruit softening and under the treatments of ethylene and/or 1-MCP, a small subset of PG genes highly linked to the softening process was identified. In natural ripening fruit of QJB, nine PGs (PpPG1, -10, -12, -13, -15, -23, -21, -22 and -25) exhibited softening-associated patterns and up-regulated expression during storage. Among these, eight PGs (all except PpPG25), which displayed dramatic up-regulation after ethylene treatment and significant down-regulation after 1-MCP treatment (Figure 7), emerge as strong candidates to play a role in QJB fruit softening. However, in natural slowly-softening QW fruit, only PpPG15, -21 and -22 expression showed softening-associated patterns during storage, and also displayed dramatic up-regulation after ethylene treatment (Figure 7), which indicated that PpPG15, -21 and -22 play a role in QW fruit softening. Thus, as common candidate genes in two varieties, PpPG15, -21 and -22 may play major roles in peach fruit softening. Moreover, the number of PG family members were involved in fruit softening was different in cultivars having different softening characteristics. More PG members play a role in fast softening peach varieties, and fewer PG members are active in slow-softening peach fruit.
Ethylene has been assigned a major role in the ripening of climacteric fruits, controlling the transcription of softening-related genes [53]. Hiwasa et al. (2003) reported that ethylene is required for PG expression even in the late ripening stage [15]. A stony hard type peach fruit characterized by a lack of ethylene production and firm flesh in the mature fruit retained low PpPG mRNA expression and endoPG enzyme activity during storage. Ethylene treatment resulted in stony hard type peach fruit softening rapidly and increasing PpPG mRNA expression and endo-PG enzyme activity [54,55]. Our results also showed that endo-PG and exo-PG enzyme activity and expression of most PG members is induced by exogenous ethylene treatment, but there are differences between two cultivars. PpPG11, -14 and -35 showed down-regulated or stable expression in naturally ripening QJB and QW fruit during storage, and their expression was not affected by ethylene treatment during QJB storage. However, in QW fruit, the expression of PpPG11, -14 and -35 increased sharply after ethylene treatment (Figure 7), which suggests that PpPG11, -14 and -35 do not play a role in softening of QJB fruit, but participate in ethylene-induced ripening and softening in QW fruit.
In addition, the expression of PpPG1, -10 and -12 remained at a lower level in QW fruit during natural ripening, but increased dramatically throughout the storage period when treated with ethylene. This was accompanied by a rapid increase in ethylene evolution (Figure 7), which suggests that PpPG1, -10 and -12 rarely affect normal softening of QW peach fruit, but when stimulated with ethylene, their expression levels rapidly increase and accelerate the softening of QW peach fruit. The expression levels of PpPG24, -25 and -38 were high and stable during the whole storage period, and were hardly affected by ethylene treatment in either QW or QJB variety fruit (Figure 7), which suggests that the expression of these PGs is not regulated by ethylene, and not related to peach fruit softening.

Plant Materials and Treatment
Peach fruit (Prunus persica L. Batsch cv. "Qian jian bai" (QJB) and "Qin wang" (QW)) from the Experimental Station of the College of Horticulture, Northwest A&F University in Yangling, Shaanxi, China, were used in this study. In our experiment, flesh texture trait of QW, whose phenotype resembles very much the stony hard type flesh in firmness and crispness, but when fully ripe becomes melting and releases ethylene. This flesh texture trait results in remarkable keeping quality, particularly on the tree, and is rewarding for growers, consumers and researchers [56].
Fruit with no visible defects were harvested when they were commercially mature (fading peel color that is light green, partially red; slightly hard flesh). For QJB, all of the intact fruit were divided randomly into three groups. One group was soaked in 1 g/L ethephon for 15 min at 25 ± 1 • C. The second group was treated with 5 µL/L 1-methylcyclopropene (1-MCP) for 24 h in a closed container, and the third group was soaked in water for 15 min and served as the control group. For QW, fruit were divided into two groups. One group was soaked in 1 g/L ethephon for 15 min at 25 ± 1 • C. The other group was soaked in water and served as the control group. Each group consisted of 150 fruit, which were stored at 25 • C. Samples were taken every other day until the flesh fully softened. All samples were frozen with liquid nitrogen and stored at −80 • C.

RNA Extraction and Reverse Transcription
Total RNA was extracted according to the protocol described by Lester et al. (1994) [35]. RNA integrity and quality were tested using an ultraviolet spectrophotometer (Thermo NanoDrop 2000, Wilmington, DE, USA) and electrophoresis in 1% agarose gels, respectively. Reverse transcription was completed using the PrimeScript RT Reagent Kit with gDNA Eraser (Takara, Beijing, China).

Identification of Peach PG Family Members
All members of the peach PG family were identified using two steps. First, hidden Markov model (HMM) profiles of the PG domain (Accession No. PF00295) were retrieved from the Pfam database [57]. Second, these domains and identified Arabidopsis PG sequences [12] were used as queries to search the GenBank non-redundant protein database and the Peach Genome Database [58]. Then all of the candidate peach PG family members were submitted to SMART [59] and Pfam to determine the presence of the domain. Those candidate PG family members that contained at least two highly conserved domains (of domains I, II, III, and IV) of PGs [60], were considered to be PGs.

Multiple Sequence Alignment, Phylogenetic Analysis, Exon/Intron Structure Determination and Chromosome Locations
Multiple sequence alignment was performed and conserved domains were analyzed using DNAMAN 6.0 with default settings. A phylogenetic tree was generated with MEGA 6.0 using the neighbor-joining (NJ) method, and edge support was estimated using 1000 bootstrap replicates. Exon/intron structure was generated with the online software GSDS [61]. Consensus sequence graphs were obtained with the online software WebLogo [62]. Signal peptides were analyzed using SignalP v3.0 [63]. Molecular weights and pIs of the proteins were calculated with the ExPASY Compute pI/M W tool [64]. To identify conserved motifs within the PG genes, MEME [65] motif analysis was performed. For chromosome locations, data were downloaded from NCBI [66] and genes were mapped to the chromosomes with MapDraw [67].

RNA Deep Sequencing and Library Construction
The QJB fruits from 0, 2, and 4 days after harvest were selected to extracted RNA. The quantity and quality of extracted RNA were tested using using a NanoDrop ND1000 spectro-photometer (Thermo NanoDrop 2000, Wilmington, DE, USA) and electrophoresis in 1% agarose gels, respectively. A~2-µg total RNA from each sample was used for constructing cDNA libraries and they were sequenced using an Illumina HiSeq TM 2000 by the Biomarker Biotechnology Corporation (Beijing, China). The experiment was repeated three times. The detailed methods of raw fastq files handling from RNA-seq were carried out according to Xing [68].

Real-Time Quantitative PCR (qPCR) Assays
Real-time quantitative PCR (qPCR) was performed using an iQ5 real-time PCR system (BioRad, Plano, TX, USA). Gene-specific primers were designed using Beacon Designer 8.0 (Table S2). For each sample, 1 µL cDNA, 1 µL of each primer, 2 µL ddH 2 O and 5 µL of 2× SYBR Premix ExTaq II (TaKaRa, Dalian, China) were used for qPCR in a total volume of 10 µL. The qPCR protocol was as specified for the SYBR Premix Ex Taq kit, 1 min at 95 • C, followed by 40 cycles of 15 s at 95 • C, 20 s at annealing temperature and 20 s at 72 • C, followed by 10 s at 95 • C, followed by 40 cycles to construct a melting curve. Peach 18S ribosomal RNA (18S rRNA) was selected as the reference gene. Relative expression level data were analyzed using the 2 −∆∆Ct method [69]. Each sample was analyzed in triplicate.

Flesh Firmness, Ethylene Production and Determination of Enzyme Activity
Firmness was measured using the GY-4 firmness meter (Top instrument Co., Ltd., Hangzhou, China) equipped with a 7.9-mm probe. In each group, six fruit were randomly selected and a small piece of epicarp was peeled off from two symmetrical positions on each fruit to attach the probe. Ethylene production was determined as described by Liguori et al. (2004) [70]. Nine fruit were weighed and sealed in a jar for 1 h. An air sample (1 mL) from the jar was analyzed using a gas chromatograph (Trace GC Ultra, Thermo Fisher, New York, NY, USA). The activities of exo-PG and endo-PG were measured according to Cao et al. (2014) [71].

Statistical Analysis
Microsoft Excel 2010 and DPS v7.05 were used for data processing and correlation analysis, and SigmaPlot 10.0 was used to prepare figures.

Conclusions
In conclusion, a genome-wide analysis of PG gene family in peach was performed to reveal gene structure, phylogenetic relationship, and expression profiles during fruit softening. During the normal ripening, eight PGs (PpPG1, -10, -12, -13, -15, -23, -21, and -22) in fast-softening QJB fruit and three PGs (PpPG15, -21, and -22) in slow-softening QW fruit exhibited softening-associated patterns, which were also affected by ethylene treatment. Our results suggest that the different softening characters in QW and QJB fruit are related to the amount of PG members.