Comparative transcriptome analyses of fruit development among pears, peaches, and strawberries provide new insights into single sigmoid patterns

Background Pear fruit exhibit a single sigmoid pattern during development, while peach and strawberry fruits exhibit a double sigmoid pattern. However, little is known about the differences between these two patterns. Results In this study, fruit weights were measured and paraffin sections were made from fruitlet to maturated pear, peach, and strawberry samples. Results revealed that both single and double sigmoid patterns resulted from cell expansion, but not cell division. Comparative transcriptome analyses were conducted among pear, peach, and strawberry fruits at five fruit enlargement stages. Comparing the genes involved in these intervals among peaches and strawberries, 836 genes were found to be associated with all three fruit enlargement stages in pears (Model I). Of these genes, 25 were located within the quantitative trait locus (QTL) regions related to fruit weight and 90 were involved in cell development. Moreover, 649 genes were associated with the middle enlargement stage, but not early or late enlargement in pears (Model II). Additionally, 22 genes were located within the QTL regions related to fruit weight and 63 were involved in cell development. Lastly, dual-luciferase assays revealed that the screened bHLH transcription factors induced the expression of cell expansion-related genes, suggesting that the two models explain the single sigmoid pattern. Conclusions Single sigmoid patterns are coordinately mediated by Models I and II, thus, a potential gene regulation network for the single sigmoid pattern was proposed. These results enhance our understanding of the molecular regulation of fruit size in Rosaceae.


Background
Within the Rosaceae, different fleshy fruit types showed two different fruit growth patterns. Pome fruits such as apple and pear exhibit a single sigmoid pattern in which fruits undergo extensive cell division during the first few weeks immediately following fertilization, after which almost all growth is due to cell enlargement [1][2][3][4]. Stone fruits such as late-maturing peach exhibit a double sigmoid pattern in which two rapid-growth stages are separated by a slow-growth stage [2,5,6]. Interestingly, strawberry fruits exhibit either a single or double sigmoid pattern in different varieties [7][8][9][10][11][12]. These results indicate that fruit growth patterns are determined by other factors than by the type of fruit.
Single and double sigmoid patterns have been reported in the last century in Rosaceae fruit species [2], but the difference between these patterns has not been explored at the molecular level until now. In Rosaceae, pear is a pome fruit and was selected for studies of single sigmoid pattern, while peach is a drupe fruit that presents a double sigmoid pattern and was selected as a control. Moreover, strawberry is an aggregate fruit that present sigmoid or double pattern and was selected as another control. The fruits in these three species were selected and performed for transcriptome sequencing to explore the difference between single and double patterns. First, developing and matured fruits were collected for measuring fruit weight and calculating cell number to clarify the periods of rapid enlargement and intervals in peaches and strawberries. Based on the completed genome projects of pears, peaches, and strawberries [46][47][48], transcriptome sequencing was conducted to screen for differentially expressed genes (DEGs) related to fruit enlargement in the three Rosaceae species. Validation of the candidate genes was conducted based on reported QTLs [35,39,44], functional annotation was also performed. Moreover, a dual-luciferase assay was conducted between the screened transcription factors and cell expansion-related genes. Finally, molecular regulation of the single sigmoid pattern was discussed. This is the first study to report on the difference between single and double sigmoid patterns at the molecular level. These results will enhance our understanding of the molecular regulation of fruit size in Rosaceae.

Results
Investigation of growth curves in pear, peach, and strawberry fruits To investigate the growth curves of pear, peach, and strawberry fruits, the fruit weight of each species was measured from fruitlet to maturation. Results revealed a single sigmoid pattern growth curve in pear fruits ( Fig. 1a; Figure  S1), but a double sigmoid pattern in peach and strawberry fruits (Fig. 1a). Further analysis of paraffin section showed that the cell number of per fruit in three fruit species mainly increased before 42, 21, and 19 days after full blooming (DAFB), respectively, and then maintained stable until to fruit ripening (Fig. 1b). In contrast, the cell size changed little before 42, 21, and 19 DAFB, respectively, and then the cell expansion was happened (Fig. 2). These results revealed that pear, peach, and strawberry fruits underwent extensive cell division before 42, 21, and 19 DAFB, respectively, then underwent cell enlargement. In addition, both the single and double sigmoid curve occurred after these days, we concluded that both single and double sigmoid patterns were a result of cell expansion, not cell division [1][2][3][4].

Transcriptome sequencing and differential expression analyses
To uncover the difference between single and double sigmoid patterns, the fruit at fruitlet, first enlargement, interval, second enlargement, and maturation (Table S1) in peach and strawberry were selected for transcriptome sequencing. These five stage samples were designated as PP1 to PP5 in peaches (Prunus persica) and FA1 to FA5 in strawberries (Fragaria × ananassa) (Table S1), respectively. In pears (Pyrus bretschneideri), the fruitlet (28 DAFB), early enlargement (98 DAFB), middle enlargement (126 DAFB), late enlargement (140 DAFB), and maturation (168 DAFB) stage fruits were selected and designated as PB1 to PB5, respectively. A total of 1.41, 1.56, and 1.44 Gb raw reads were generated from pear, peach, and strawberry fruits, respectively. After removing low-quality reads, an average of 91.47, 100.01, and 92.3 Mb clean reads were used for mapping to the reference genome in pear, peach, and strawberry fruits, respectively. The average percentages of total mapped reads were 73.99, 88.64, and 72.27% in pear, peach, and strawberry fruits, respectively. A total of 31,910 genes were detected in pear fruits. Of these genes, 25,633 were commonly expressed in all tested fruits, while 1303, 231, 162, 214, and 254 were specifically expressed from PB1 to PB5, respectively ( Figure S2). A total of 27,747 genes were detected in peach fruits. Of these genes, 22,480 were commonly expressed in all tested fruits, while 652, 189, 149, 113, and 291 were specifically expressed from PP1 to PP5, respectively ( Figure S2). A total of 35,502 genes were detected in strawberry fruits. Of these genes, 27,023 were commonly expressed in all tested fruits, while 1156, 303, 265, 702, and 325 were specifically expressed from FA1 to FA5, respectively ( Figure S2).

Overlap of DEGs with QTL regions related to fruit size
Fruit size, which includes fruit weight, height, width, and depth, has been widely studied in pear, peach, and strawberry fruits. In pears, fruit size was anchored by 21 QTL regions covering 1279 genes [35][36][37], of which, 123, 72, and 117 genes were detected in PB2-, PB3-, and PB4-DEGs, respectively (Table S2). In peaches, fruit size was anchored by 36 QTL regions covering 7100 genes [39][40][41], of which, 1198 and 1001 genes were detected in PP2and PP4-DEGs, respectively (Table S2). In strawberries, fruit size was anchored by three QTL regions covering 192 genes within a 150 kb range [44,45], of which, four genes were detected in FA2-DEG (Table S2). These results revealed that the DEGs overlapped with QTLs related to fruit size, indicating that screening for fruit enlargement candidate genes is reliable based on the transcriptome analyses of these three Rosaceae species.

Conception of fruit enlargement in Rosaceae
Pear, peach, and strawberry fruits are Rosaceae species that have relatively close genetic backgrounds. However, Fig. 1 Measurements of fruit weight and cell number in pear, peach, and strawberry fruits. a Fresh weight (g) of pear, peach, and strawberry fruitlets to matured fruit samples. b Cell numbers of pear, peach, and strawberry fruitlets to matured fruits per fruit. Standard errors (SEs) were calculated from 15 biological replicates pear fruits exhibit a single sigmoid pattern, while peach and strawberry fruits exhibit a double sigmoid pattern (Fig. 1a). These growth patterns are a result of the differences between the middle enlargement stage in pears (PB3) and interval in peaches and strawberries (PP3/ FA3). In peaches, two fruit enlargement stages were mediated by PP2-and PP4-DEGs, while the interval was controlled by genes exhibiting the opposite trend in PP3-DEG compared to PP2-and PP4-DEGs. This conclusion was also observed in strawberry fruits. Therefore, it was speculated that the genes involved in the interval stages must possess correlated expression profiles with the fruit growth patterns of peaches and strawberries. In pears, the genes involved in the middle enlargement stage correlated with the expression profiles of fruit growth patterns or were independent from the genes involved in the early and late enlargement stages.
By comparing the double sigmoid pattern in peaches and strawberries, the single sigmoid pattern in pears was explained by two models. In the first model (Model I), the intervals in peach and strawberry fruits was replaced by the first and second enlargement stages, leading to the single sigmoid pattern, which was mediated by the genes associated with all three fruit enlargement stages in pears. In the second model (Model II), the intervals in peach and strawberry fruits were independent of other fruit enlargement stages, resulting in the single sigmoid pattern, which was mediated by the genes involved in the middle enlargement stage, but not the early or late enlargement stages in pears.

Identification of cell development genes in Model I
To examine the possibility of Model I, the genes that present correlated expression profiles with the two fruit growth patterns were identified in the three Rosaceae species. In peaches, 90 genes were upregulated in PP2-and PP4-DEGs, but downregulated in PP3-DEG; this gene set was designated as PPP. Meanwhile, 162 genes were downregulated in PP2-and PP4-DEGs, but upregulated in PP3-DEG; this gene set was designated as PPN. In strawberries, 43 genes were upregulated in FA2-and FA4-DEGs, but downregulated in FA3-DEG; this gene set was designated as FAP. Moreover, 26 genes were downregulated in FA2and FA4-DEGs, but upregulated in FA3-DEG; this gene set was designated as FAN. In pears, 592 genes were upregulated and 335 were downregulated, which overlapped among PB2-, PB3-, and PB4-DEGs (Fig. 4a, b), correlating with the single sigmoid pattern in pears. The up-and downregulated gene sets were designated as PBPC and PBNC, respectively. Among these 927 genes, 155 were involved in cell development, including cell wall biogenesis, cell growth, and cell division ( Figure S3A). However, the comparative analysis of the genes in PBPC, PPP, and FAP detected no orthologous genes among the three Rosaceae species, nor did the comparative analysis of the genes in PBNC, PPN, and FAN. These results were inconsistent with Model I, therefore, the genes associated with all three fruit enlargement stages in pears were different from the genes involved in the intervals in peaches and strawberries.
Although the genes in PBPC were different from the genes in PPP and FAP, it was not clear whether all the genes in PBPC were associated with fruit enlargement as the orthologous genes in peach and strawberry fruits may be included in PPN/FAN. A similar inference was made for the genes in PBNC. To verify these speculations, a comparative analysis was conducted on the genes in PBPC, PPN, and FAN. Results revealed that 44 genes in PBPC were orthologous to the genes in PPN or FAN, suggesting that these genes were not associated with the single sigmoid pattern; the remaining 548 genes in pears may be involved in fruit enlargement (Fig. 4a). Similarly, 47 genes in PBNC were orthologous to the genes in PPP or FAP, suggesting that these genes were not associated with the single sigmoid pattern; the remaining 288 genes in pears may be involved in fruit enlargement (Fig. 4b). In these 836 DEGs in pears, 90 were involved in cell development, including the cell cycle and cell wall metabolism (Table S3). Notably, of these genes, 25 were located within the fruit weight QTL regions (Table 1). Clearly, several of these genes were specifically associated with fruit enlargement in pears and the difference between the single and double sigmoid patterns may result from these genes.

Identification of cell enlargement genes in Model II
To examine the possibility of Model II, the genes that mediate the middle enlargement stage, but not early or late enlargement, were isolated in pears. As a result, 512 upregulated and 204 downregulated genes were detected in PB3-DEG; these gene sets were designated as PBPO and PBNO, respectively (Fig. 4c, d). Among these 706 genes, 22 were involved in cell wall biogenesis, ubiquitin, and auxin ( Figure S3B). However, comparative analysis of the genes in PBPO, PPP, and FAP detected no orthologous genes among the three Rosaceae species, nor did the comparative analysis of the genes in PBNO, PPN, and FAN. These results were inconsistent with Model II, therefore, the genes involved in the middle enlargement stage in pears were different from the genes involved in the intervals of peaches and strawberries.
Although the genes in PBPO were different from the genes in PPP and FAP, it was not clear whether all the genes in PBPO were associated with fruit enlargement as the orthologous genes in peach and strawberry fruits may be included in PPN/FAN. A similar inference was made for the genes in PBNO. To verify these speculations, a comparative analysis was conducted among the genes in PBPO, PPN, and FAN. Results revealed that 47 genes in PBPO were orthologous to the genes in PPN or FAN, suggesting that these genes were not associated with the single sigmoid pattern. The remaining 465 genes in pears may be involved in fruit enlargement (Fig.   4c). Similarly, 20 genes in PBNO were orthologous to the genes in PPP or FAP, suggesting that these genes were not associated with the single sigmoid pattern. The remaining 184 genes in pears may be involved in fruit enlargement (Fig. 4d). Of these 649 DEGs in pears, 63 were involved in cell development, including cell wall biogenesis, cell growth, and cell metabolism (Table S4). Notably, of these genes, 22 were located within the fruit weight QTL regions (Table 1). Clearly, several of these genes were specifically associated with fruit enlargement

Expression profiles of fruit enlargement candidate genes
To confirm the transcriptome assay, 12 candidate genes associated with cell wall, cell cycle, ubiquitination, phytohormones, cytochrome, ankyrin, transcription factor, and leucine-rich repeats (LRR) from Models I and II were selected for quantitative real-time polymerase chain reaction (qRT-PCR) analysis of pear, peach, and strawberry fruits. Results revealed that these genes had similar expression profiles as the transcriptome assay results, indicating that these findings were reliable ( Figure S4). Moreover, to test whether the candidate genes exhibited similar expression profiles in different pear cultivars, the 'Housui', 'Cuiguan', and 'Xueqing' cultivars were investigated at the same five fruit enlargement stages and analyzed by qRT-PCR ( Figure  S1). Results revealed that the expression profiles of the 12 aforementioned genes in the three cultivars were almost identical to the 'Dangshansuli' cultivar ( Fig. 5), suggesting that these candidate genes identified from 'Dangshansuli' were also involved in the middle enlargement stages of other pear cultivars. Therefore, these candidate genes may be associated with the single sigmoid pattern in pear fruits.

Regulation of bHLHs in cell expansion-related genes
In a previous study, bHLH was defined as the upstream factor regulating cell expansion-related genes in peaches [13]. In this study, four bHLH genes, bHLH3 (Pbr009044.1), bHLH30 (Pbr016145.1), bHLH106 (Pbr013890.1), and bHLH144 (Pbr039557.1), were detected in the genes associated with the single sigmoid pattern, as well as cell expansion-related genes, including cellulose synthase (CES), glycosyltransferase (GT), microtubule-associated protein (MAP), UDP-glycosyltransferase (UGT), COP9 signalosome complex subunit (CSN), exocyst complex component (EXOC), expansin (EXP), xyloglucan galactosyltransferase (XG GalT), and xyloglucan glycosyltransferase (XG GUT). To investigate the role of the four bHLH transcription factors (TFs) on the expression of cell expansion-related genes, full-length sequences of the four bHLH genes were amplified and inserted into pSAK277 as effectors. Additionally,~2000 bp sequences in the promoter of cell expansion-related genes were inserted into pGreen 0800-LUC as reporters (Fig. 6a). The dual luciferase assay results revealed that bHLH3, bHLH30, and bHLH106 enhanced the activities of firefly luciferase (LUC), which were driven by the promoters of 11 of the 14 selected genes, while bHLH144 enhanced the activities of LUC and was driven by the promoters of 10 of the 14 selected genes (Fig. 6b). Notably, LUC was driven by the promoters of cellulose synthase, glycosyltransferase-1, glycosyltransferase-2, UDP-glycosyltransferase-2, expansin, xyloglucan galactosyltransferase-1, and xyloglucan glycosyltransferase, and exhibited higher activities in tobacco leaves overexpressing any of the bHLH genes compared to tobacco leaves transformed with an empty vector. These results suggested that the four bHLH TFs interacted with most promoters of cell expansion-related genes and drove their expression.

Discussion
Potential gene regulation network for the pear single sigmoid pattern Single and double sigmoid patterns have been reported over the last century in Rosaceae fruit species [2], however, the difference between these patterns had not been explored at the molecular level until now. In this study, based on the investigated physiological phenotypes, two models of fruit enlargement in pears were proposed. By comparative transcriptome analyses of the selected fruits in three Rosaceae fruit species, several genes associated with the cell cycle [19][20][21], polysaccharides and cell development [22][23][24][49][50][51][52][53][54][55][56], were detected in Models I and II (Tables S3  and S4). Therefore, models I and II play important roles in generating the single sigmoid pattern.
In addition to the above genes, several transcription factors were also detected in Models I and II (Table 1; Tables  S3 and S4), including zinc finger proteins (ZFPs), which control cell size during plant organogenesis [57], and bHLHs, which regulate cell extension by transducing auxin signaling [27]. bZIP mediates cell expansion by affecting the biosynthesis and signal transduction of gibberellins and auxins [24,58], while ankyrin repeat proteins enhance auxin biosynthesis via bZIP [59]. WRKY and NAC are involved in cell development [30,60,61], as well as auxin biosynthesis [62,63], while proliferating cell factor (TCP) and MYB are only associated with auxin biosynthesis [64,65]. Clearly, auxin signal transduction may be involved in the middle enlargement stage of pear fruits. In this study, auxin-induced factors, including auxin response factor, auxin-induced proteins, and auxinresponsive proteins, were detected in Models I and II, which are associated with cell expansion [66,67]. Moreover, the detected cytochrome p450 proteins stimulate plant organ growth by increasing cell size [17]. The transcriptional activation of the four bHLHs in 14 candidate genes associated with the pear single sigmoid pattern of the two models was also confirmed by the dual-luciferase assay. These results revealed that the promoters of the 14 genes were activated by bHLHs (Fig. 6). Based on these results, a potential gene regulation network was proposed for the pear single sigmoid pattern (Fig. 7). This network indicates that phytohormone auxin was regulated by MYB, WRKY, TCP, NAC, and ankyrin-mediated bZIP, which thereby triggered the expression of bHLH and auxin-induced factors. Simultaneously, bZIP triggered gibberellin signal transduction to activate gibberellin-responsive proteins. Auxininduced factors along with bHLH, ZFP, cytochrome P450, and gibberellin-responsive proteins promoted the expression of beta-galactosidase, cellulose synthase, GDSL esterase/lipase, Glycosyltransferase, microtubule-associated protein, COP9 signalosome complex subunit, exocyst complex component SEC15A, expansin, and xyloglucan galactosyltransferase to promote cell expansion, which resulted in the middle enlargement of pear fruits.

Positive selection was observed in pear and peach fruit enlargement genes
Plant domestication is a long-term, people-based plant interaction that favors the alleles of genes that control certain traits of interest in increased frequency [68]. In fleshy fruits, fruit enlargement is a typical phenotype observed during fruit development and maturation. In a previous study, tomato fruits exhibited a single sigmoid pattern during fruit growth [2], and fruit mass was domesticated from the small fruits of their ancestors to the big fruits of modern tomatoes [55,69]. The similar domestication of fruit size was detected in pears [70], peaches [71,72], apples [73] and sweet cherries [43].
In this study, the dynamic weight of developing and mature pear, peach, and strawberry fruits was measured. Results confirmed that pear fruits exhibited a single sigmoid pattern during fruit growth, while peach and strawberry fruits exhibited a double sigmoid pattern. However, it is unclear whether the fruit growth patterns of these three Rosaceae species undergo positive selection. Previously, the selective sweep regions of fruit size were isolated in peach and pear fruits [70,72,74]. In peaches, 707 and 507 genes correlated with the first and second enlargement stages, respectively, and were located within the selective sweep regions. Moreover, 395 genes correlated with the first and second enlargement stages [71,72]. These results indicate that the first and second enlargement stages were domesticated during the evolution of these species (Table S5). Similarly, in pears, 66, 32, and 74 genes correlated with the early, middle, and late enlargement stages, respectively, and were located within the selective sweep regions. Moreover, 21 genes correlated with the three fruit enlargement stages (Table S5). These results suggest that the three fruit enlargement stages were domesticated during Fig. 7 Potential gene regulation network of the pear single sigmoid pattern showing TCP, NAC, MYB, WRKY, and ankyrin-mediated bZIP-regulated auxin expression. bZIP triggered gibberellin signal transduction to activate gibberellin-responsive proteins. Auxin-induced factors along with bHLH, ZFP, cytochrome P450, and gibberellin-responsive proteins promoted the expression of genes related to cell expansion, which resulted in the middle enlargement of pear fruits the evolution of these species. Collectively, these findings indicate that each fruit enlargement stage was domesticated in pear and peach fruits.
In previous studies, the double sigmoid pattern of peach fruits was explained by two hypotheses: the competition for assimilates between the seed and pericarp, and the hormonal control of pericarp growth by the seed [6]. In this study, these hypotheses may co-exist in the three fleshy fruits. First, strawberry fruit development is dependent on seed fertilization-induced hormones [75], indicating that hormones are necessary for fruit enlargement. Theoretically, hormone-controlled fruit enlargement should not be interrupted. However, when strawberry fruits were bleaker in color ( Figure S5), both hormones and assimilates were used for anthocyanin biosynthesis or fruit enlargement, leading to slow fruit growth. Similarly, when peach fruits exhibited stone hardening ( Figure S5), both hormones and assimilates were provided for endocarp lignification or fruit enlargement, inducing slow fruit growth. In contrast, due to the lack of dramatic changes observed during pear fruit development (Fig. 1a), both hormones and assimilates were used for fruit enlargement, which resulted in the single sigmoid pattern.

Conclusions
In this study, pear fruits exhibited a single sigmoid pattern, while peach and strawberry fruits exhibited a double sigmoid pattern. These patterns resulted from cell expansion, not cell division. By comparative transcriptome analysis among pear, peach, and strawberry fruits, 836 genes were found to be associated with all three fruit enlargement stages in pear fruits (Model I), while 649 genes were associated with middle enlargement stage, not early or late enlargement (Model II). Therefore, the pear single sigmoid pattern appeared to be coordinately mediated by these proposed models. Interestingly, most genes in these models were annotated and correlated with cell development. Moreover, 47 genes were located within the QTL regions related to fruit weight. Based on previous reports and the dualluciferase assay, a potential gene network of the single sigmoid pattern was drafted in this study (Fig. 7).
Peach fruits were collected every 7 days from 14 DAFB to maturation. The 'Hongyan' strawberry cultivar (F. × ananassa) was grown in a greenhouse at Zhen Jiang Agricultural Academy, China. A total of 300 flowers were selected and marked at the initial flowering stage; the fruits derived from these flowers were collected every 4 days from three DAFB to maturation. Fruit weight was measured using 15 fruits from each period for all fruit species.
Moreover, to analyze the expression profiles of whole predicted genes during fruit swelling in pear, peach, and strawberry fruits, each collection had a minimum of three independent replicates. The collected pear and peach fruits were snap frozen in liquid nitrogen and ground into a fine powder. Similarly, after removing the seeds from the surface of strawberry fruits, the remaining tissues were mashed in liquid nitrogen and stored at − 80°C for further analysis.

Library preparation and sequencing
Total RNA was extracted using an RNAprep Pure Plant kit (Tiangen Biotech, Beijing, China). RNA degradation and contamination were monitored on 1% agarose gel. RNA purity was checked using a NanoPhotometer® spectrophotometer (Implen, CA, USA). RNA concentrations were measured using a Qubit® RNA Assay kit in a Qubit® 2.0 flurometer (Life Technologies, CA, USA). RNA integrity was assessed using an RNA Nano 6000 Assay kit on a Bioanalyzer 2100 system (Agilent Technologies, CA, USA). The RNA integrity number (RIN) of pear, peach, and strawberry fruits was > 8.8, which met the requirement for building libraries. After qualitative and quantitative detections of total RNA, ribosomal RNA (rRNA) was depleted using an Epicentre Ribo-zero™ rRNA Removal kit (Epicentre, WI, USA). rRNA-deleted RNA fragmentation was used to synthesize first and second cDNA strands with a random hexamer primer in NEBNext First Strand Synthesis Reaction buffer (5×). Then, the NEBNext adaptor was added to adenylated DNA fragments followed by purification and screening of cDNA fragments, which were 150-200 bp in length. A tru-Seq PP Cluster kit v3-cBot-HS (Illumina, CA, USA) was used to perform clustering. Then, the libraries for 'Danshansuli', 'Liangyuan', and 'Hongyan' were sequenced on an Illumina Hiseq 4000 platform (Illumina, CA, USA) with three replicates.

Chromosome location of DEGs and orthologous analysis
The location of DEGs was retrieved based on mRNA information. DEGs were located within the reported QTL regions were statistically analyzed in python script (https://github.com/Peims/Calculate-the-distance-between-the-gene-and-the-marker). An orthologous genes analysis was conducted using OrthoFinder v2.3.3 [82].

qRT-PCR
Total RNA was extracted using a Plant Total RNA Isolation kit plus (Gore Gene, Chengdu, China). The first strand of cDNA was synthesized using FastKing gDNA Dispelling RT SuperMix kits (Tiangen Biotech, Beijing, China), which was subsequently used as a template for qRT-PCR. qRT-PCR was performed on a Lightcycle-480 system using LightCycler 480 SYBR Green I Master mix (Roche, Basel, Sweden). The reaction mixture was conducted as previously reported [74]. The gene-specific primers used in this study are provided (Table S6). All PCR experiments were performed using three independent biological samples, which included three technical replicates. The pear SNF, peach TEF, and strawberry DBP genes were used as normalizers. An independent sample t-test was used to analyze the qRT-PCR results between each sample. A correlation analysis between transcriptome and qRT-PCR results was conducted in python script (https://github.com/Peims/ calculate-the-pearsoner).

Vector constructs and dual luciferase assays
To detect the transcriptional activation of four bHLHs on 14 candidate genes associated with the single sigmoid pattern of Models I and II, full-length sequences of the four bHLHs were amplified using specific primers (Table S7). Phanta® Max Super-Fidelity DNA polymerase (P505, Vazyme Biotech Co., Ltd., China) was introduced into the pSAK277 vector to construct the effectors (Fig. 6a). The2 000 bp promoter sequence of the 14 genes was inserted into the pGreen 0800-LUC vector to construct reporters. Then, the correctly recombinant plasmids were transformed into the Agrobacterium strain, GV3101, using pSoup for pGreen 0800-LUC vector transformation. The GV3101 strain contained an effector and reporter was coinfiltrated into tobacco leaves [83]. The empty pSAK277 transient expression vector was used as the negative control. For each promoter-TF comparison, at least six infiltrated leaves using the same Agrobacterium culture were used in the dual-luciferase assay. LUC and Renillia luciferase (REN) activity were assayed using a Dual-Luciferase® Reporter Assay system (Promega, WI, USA) following the manufacturer's instructions. Relative LUC activities were calculated as the ratio of LUC/REN for each comparison and normalized to the control in each experiment. Statistical analyses were conducted using SPSS v17.0 software (SPSS Inc., CHI, USA). Student's t-tests were performed to determine significant differences (p < 0.05).