Transcriptome Analysis Reveals the Role of GA3 in Regulating the Asynchronism of Floral Bud Differentiation and Development in Heterodichogamous Cyclocarya paliurus (Batal.) Iljinskaja

Cyclocarya paliurus is an important medical plant owing to the diverse bioactive compounds in its leaves. However, the heterodichogamy with female and male functions segregation within protandry (PA) or protogyny (PG) may greatly affect seed quality and its plantations for medicinal use. To speculate on the factor playing the dominant role in regulating heterodichogamy in C. paliurus, based on phenotypic observations, our study performed a multi comparison transcriptome analysis on female and male buds (PG and PA types) using RNA-seq. For the female and male bud comparisons, a total of 6753 differentially expressed genes (DEGs) were detected. In addition, functional analysis revealed that these DEGs were significantly enriched in floral development, hormone, and GA-related pathways. As the dominant hormones responsible for floral differentiation and development, gibberellins (GAs) in floral buds from PG and PA types were quantified using HPLC-MS. Among the tested GAs, GA3 positively regulated the physiological differentiation (S0) and germination (S2) of floral buds. The dynamic changes of GA3 content and floral morphological features were consistent with the expression levels of GA-related genes. Divergences of GA3 contents at S0 triggered the asynchronism of physiological differentiation between male and female buds of intramorphs (PA-M vs. PA-F and PG-F vs. PG-M). A significant difference in GA3 content enlarged this asynchronism at S2. Thus, we speculate that GA3 plays the dominant role in the formation of heterodichogamy in C. paliurus. Meanwhile, the expression patterns of GA-related DEGs, including CPS, KO, GA20ox, GA2OX, GID1, and DELLA genes, which play central roles in regulating flower development, coincided with heterodichogamous characteristics. These results support our speculations well, which should be further confirmed.


Introduction
Cyclocarya paliurus (Batal.) Iljinskaja is the sole species in the genus Cyclocarya Iljinskaja (Juglandaceae) [1]. It is mainly distributed at an altitude of 420-2500 m in subtropical mountainous areas of Jiangxi, Zhejiang, Anhui, Fujian, Hubei, Sichuan, Guizhou, Guangxi, etc. [2]. C. paliurus is generally known as the sweet tea tree, and its leaves are used to cure fever, detoxify, and alleviate pain [3]. Moreover, it was widely used in ancient China to make a medicinal tea [4]. Recently, various studies have found that many potentially bioactive compounds in the leaves of C. paliurus, including polysaccharides, proteins, minerals, phenolic compounds, vitamins, saponins, sugars, triterpenoids, and amino acid have beneficial effects on antihypertensive activity, hypoglycemic activity, hypolipidemic activity, and antioxidant activity [5,6]. There is an urgent need to cultivate C. paliurus for leaf production to fulfill the need for its medicinal use. Different stages of floral buds in C. paliurus were selected for RNA-Seq analysis based on morphological changing. At S0, all flower buds were not visible. At S1, although PA-F buds enclosed by leaf buds were not visible, male and PG-F buds were clearly differentiated. At S2, all the flower buds began to protrude, except PA-F. At S3, feathery stigma formed in PG-F, while this was not observed in PA-F; PA-M turned from turquoise to yellow-green with anther dehiscence, while PG-M still remained turquoise. At S4, the feathery stigma of PG-F opened and the mature pollen of PA-M was released from the dehisced anther; the stigma of PA-F was not fully formed, and the anther of PG-M was slowly dehiscent (Figure 1).

Sequencing, Assembly, and Annotation of C. paliurus
RNA-seq was performed on the Illumina Novaseq platform. Four flower buds (PG-F, PG-M, PA-F, and PA-M) at five stages (S0, S1, S2, S3, and S4) and leaves were used for transcriptome profiling. In total, more than 64 million raw reads were obtained. After quality evaluation and filtration, the number of clean reads ranged from 64,141,024 to 79,962,072 with an average of 73,855,021 and the average value of Q30 was 94.81%, indicating that the clean reads were of high quality. After alignment with the rubber tree genome sequence, more than 94% of the clean reads were mapped to the genome and pairedend reads were aligned to 35,221 C. paliurus annotated gene models (Table S1).

Identification of Differentially Expressed Genes (DEGs) during Floral Bud Differentiation and Development
Two comparisons of three floral buds (PA-M vs. PA-F, PG-F vs. PG-M) detected a total of 6753 DEGs that were consistently upregulated or downregulated in at least two 2.2. Sequencing, Assembly, and Annotation of C. paliurus RNA-seq was performed on the Illumina Novaseq platform. Four flower buds (PG-F, PG-M, PA-F, and PA-M) at five stages (S0, S1, S2, S3, and S4) and leaves were used for transcriptome profiling. In total, more than 64 million raw reads were obtained. After quality evaluation and filtration, the number of clean reads ranged from 64,141,024 to 79,962,072 with an average of 73,855,021 and the average value of Q30 was 94.81%, indicating that the clean reads were of high quality. After alignment with the rubber tree genome sequence, more than 94% of the clean reads were mapped to the genome and paired-end reads were aligned to 35,221 C. paliurus annotated gene models (Table S1).

Identification of Differentially Expressed Genes (DEGs) during Floral Bud Differentiation and Development
Two comparisons of three floral buds (PA-M vs. PA-F, PG-F vs. PG-M) detected a total of 6753 DEGs that were consistently upregulated or downregulated in at least two stages ( Figure 2). Comparatively, the number of DEGs in PA-M vs. PA-F (6170) were more than in PG-F vs. PG-M (1880) (Figure 2).

GO Functional Annotation and KEGG Pathway Enrichment of DEGs
For the PG-F vs. PG-M comparison, GO functional annotation showed that most of the DEGs during the five stages were enriched in a series of biological processes involved in flower and organ development, hormone biosynthetic and metabolic processes, and GA signaling pathways (Figure 3a). Moreover, KEGG pathway enrichment revealed plant hormone signal transduction, environmental information processing, signal transduction, cytochrome P450, and the metabolism of terpenoids (Figure 3b). Notably, many of the up-regulated DEGs (371/1880 in PG-F) were related to hormone (GA) biosynthesis and signaling pathways.

GO Functional Annotation and KEGG Pathway Enrichment of DEGs
For the PG-F vs. PG-M comparison, GO functional annotation showed that most of the DEGs during the five stages were enriched in a series of biological processes involved in flower and organ development, hormone biosynthetic and metabolic processes, and GA signaling pathways ( Figure 3a). Moreover, KEGG pathway enrichment revealed plant hormone signal transduction, environmental information processing, signal transduction, cytochrome P450, and the metabolism of terpenoids ( Figure 3b). Notably, many of the upregulated DEGs (371/1880 in PG-F) were related to hormone (GA) biosynthesis and signaling pathways.
For the PA-M vs. PA-F comparison, the functional analysis revealed that these DEGs were also enriched in floral development, hormone, and GA-related pathways ( Figure  3c,d). Similarly, a large number of up-regulated DEGs (893/6170 in PA-M) were associated with hormone biosynthesis and metabolic processes and signal transduction. These results indicated that hormones, especially GA, may play an important role in contributing to the heterodichogamy morphs with asynchronous flowering.

DEGs Related to Gibberellin in Biosynthesis and Regulatory Pathways
DEGs in two comparisons were both significantly enriched in hormone and gibberellin biosynthesis, metabolic, and signaling pathways. Moreover, a series of ABCE module genes, such as AP3, PI, AG, and SEP, could be effectively activated by GA and play key role in regulating floral stamen and pistil morphogenesis [36,37]. Hence, we especially focused upon the DEGs associated with gibberellin (GA) biosynthesis and signal pathways.
For the PG-F vs. PG-M comparison, many genes that are involved in GA biosynthesis, such as CPS (ent-copalyl diphosphate synthase), KO (ent-kaurene oxidase), GA20ox (GA 20-oxidase), GA2ox (GA 2-oxidase), and GID1 (gibberellin receptor), were highly expressed in PG-M at the S2 stage. However, DELLA genes, which negatively modulate GA signaling via binding to GAs and play important functions in inhibiting vegetative and For the PA-M vs. PA-F comparison, the functional analysis revealed that these DEGs were also enriched in floral development, hormone, and GA-related pathways (Figure 3c,d). Similarly, a large number of up-regulated DEGs (893/6170 in PA-M) were associated with hormone biosynthesis and metabolic processes and signal transduction. These results indicated that hormones, especially GA, may play an important role in contributing to the heterodichogamy morphs with asynchronous flowering.

DEGs Related to Gibberellin in Biosynthesis and Regulatory Pathways
DEGs in two comparisons were both significantly enriched in hormone and gibberellin biosynthesis, metabolic, and signaling pathways. Moreover, a series of ABCE module genes, such as AP3, PI, AG, and SEP, could be effectively activated by GA and play key role in regulating floral stamen and pistil morphogenesis [36,37]. Hence, we especially focused upon the DEGs associated with gibberellin (GA) biosynthesis and signal pathways.
For the PG-F vs. PG-M comparison, many genes that are involved in GA biosynthesis, such as CPS (ent-copalyl diphosphate synthase), KO (ent-kaurene oxidase), GA20ox (GA 20-oxidase), GA2ox (GA 2-oxidase), and GID1 (gibberellin receptor), were highly expressed in PG-M at the S2 stage. However, DELLA genes, which negatively modulate GA signaling via binding to GAs and play important functions in inhibiting vegetative and reproductive growth, were lowly expressed in PG-F at S2 (Figure 4).   For the PA-M vs. PA-F comparison, GA2ox and DELLA genes were highly expressed in PA-M at S2, but lowly expressed at S0. However, a series of genes including CPS, KO, KAO (ent-kaurenoic acid oxidase), and GA3ox (GA 3-oxidase) were highly expressed in PA-F at S2 (Figure 4). Taken together, the heterodichogamy within PA or PG might be associated with the floral induction pathway promoted by GA signaling.

GA 3 Plays a Critical Role in Floral Differentiation and Development
The effect of endogenous GAs on floral differentiation and development was investigated; the evolving patterns of four GAs (bioactive GA 1 , GA 3 , GA 4 and GA 7 ) contents are depicted in Table 1. The analysis of variance showed that there was a significant difference in the contents of GAs during the process of male and female flower differentiation and development in two mating types (p = 0.049). Among the four GA types, only small contents of GA 1 , GA 4 , and GA 7 were detected, and their fluctuations changed little across the floral developmental process. Comparatively, the contents of GA 4 and GA 7 in floral buds were negligible, although the content of GA 1 was higher than that of GA 4 and GA 7 . However, the maximum of GA 3 in four floral buds was much higher than that of the other three GA types, varying from 40.37 ng/g FW to 116.50 ng/g FW at the S0 and S2 stages, with the minimum decreasing abruptly to trace amounts at the S3 and S4 stages (Table 1). So, the ranges of GA 3 , varying from 35.81 ng/g FW to 113.97 ng/g FW, fluctuated greatly across the five stages, indicating its prominent effects on regulating floral differentiation and development. In summary, the flower bud differentiation and development in C. paliurus greatly depend on the content changes of GA 3 rather than GA 1 , GA 4 , and GA 7 . Therefore, the dynamics of GA 3 content were separately analyzed in the following study. For female floral buds (PG-F), GA 3 contents were higher at both S0 and S2 than other stages; particularly at S0, the content reached the largest value (80.46 ng/g FW), while the lowest level (<16.09 ng/g FW) was measured at other stages (S1, S3, and S4). For male floral buds (PG-M), a similar changing pattern of GA 3 content to that in PG-F was observed, with higher values at S0 and S2 but lower values at S1, S3, and S4. Remarkably, the contents in PG-M were higher at S0 (106.33 ng/g FW) and S2 (89.53 ng/g FW) than in PG-F at those stages ( Figure 5a). These results indicated that GA 3 could positively regulate the physiological differentiation (S0) and germination of male and female floral buds (S2) in PG, and make no difference at other phases; this was also supported by the significant differences of GA 3 contents in male/female flower buds over various stages (p < 0.05). Multiple comparisons revealed no significant differences between males and females at S0, suggesting that GA 3 could regulate both sexual floral physiological differentiations. However, the difference of GA 3 content between two floral buds at S2 was significant (p < 0.05), showing that a higher GA 3 content is needed for the germination of male floral buds than for female ones. All findings were confirmed by their corresponding morphological differences: differentiated male floral buds were visible but differentiated female buds were invisible at S0. Moreover, the onset of germination of male buds was earlier and at a faster rate than that of females at S2 ( Figure 5). ever, the difference of GA3 content between two floral buds at S2 was significant (p < 0.05), showing that a higher GA3 content is needed for the germination of male floral buds than for female ones. All findings were confirmed by their corresponding morphological differences: differentiated male floral buds were visible but differentiated female buds were invisible at S0. Moreover, the onset of germination of male buds was earlier and at a faster rate than that of females at S2 ( Figure 5).

PA Mating Type
In PA, GA3 content in male and female buds at all stages displayed identical patterns to those in PG, except for a lower value in PA-F at S0. These dynamics correlated well with the characteristics of sexual floral bud differentiation and growth in PA, where the process of male floral buds was prior to that of female buds (Figure 5b). The low GA3 content of female floral buds at S0 was entirely consistent with that, as their physiological distinction in S0 was not completed until S2 in the following year (Figure 1). At the S0 and S2 stages, GA3 variation in two sexual floral buds confirmed that the quantity of GA3 could determine the differentiation in floral buds and germination. Also, the higher GA3 content in male than in female buds further suggests that the germination of male floral buds could be more dependent on GA3 content than female ones.

PA Mating Type
In PA, GA 3 content in male and female buds at all stages displayed identical patterns to those in PG, except for a lower value in PA-F at S0. These dynamics correlated well with the characteristics of sexual floral bud differentiation and growth in PA, where the process of male floral buds was prior to that of female buds (Figure 5b). The low GA 3 content of female floral buds at S0 was entirely consistent with that, as their physiological distinction in S0 was not completed until S2 in the following year (Figure 1). At the S0 and S2 stages, GA 3 variation in two sexual floral buds confirmed that the quantity of GA 3 could determine the differentiation in floral buds and germination. Also, the higher GA 3 content in male than in female buds further suggests that the germination of male floral buds could be more dependent on GA 3 content than female ones. Likewise, the low GA 3 content of all leaves around male and female flower buds in both mating types indicates that the leaves may not be the source of GA 3 for flower bud differentiation and growth ( Figure 5).

Discussion
Heterodichogamy is described as the presence of two structures that exhibit male female characteristics at different times (within the same flower). Particularly, the PA is male prior to female and the PG type is female prior to male, and two flowering st appear during the flowering season. Hence, flowering asynchronization emerges not between the male and female flower of intramorphs (such as PA-F and PA-M), but the same sexual flower of intermorphs (such as PG-F and PA-F). This phenomenon been noticed in many heterodichogamous plants, particularly in Juglandaceae spe such as Juglans regia L. [35], J. mandshurica [38], Carya illinoinensis [39], Platycarya strobi [11], and C. paliurus [12]. However, our study tries to shed light on the puzzle of w regulates floral buds to realize heterodichogamy in these plants.
For male and female flowers of intramorphs, GA-related genes may regulate asynchronization by perceiving the position differences of two sexual buds. As state Mao et al. [12], in C. paliurus female buds grow at the apex of the growing shoot and m buds locate at the lateral short branch; the expression of GA-related DEGs in male female buds may be triggered under the genetic control depending on the mating t

Discussion
Heterodichogamy is described as the presence of two structures that exhibit male and female characteristics at different times (within the same flower). Particularly, the PA type is male prior to female and the PG type is female prior to male, and two flowering stages appear during the flowering season. Hence, flowering asynchronization emerges not only between the male and female flower of intramorphs (such as PA-F and PA-M), but also the same sexual flower of intermorphs (such as PG-F and PA-F). This phenomenon has been noticed in many heterodichogamous plants, particularly in Juglandaceae species, such as Juglans regia L. [35], J. mandshurica [38], Carya illinoinensis [39], Platycarya strobilacea [11], and C. paliurus [12]. However, our study tries to shed light on the puzzle of what regulates floral buds to realize heterodichogamy in these plants.
For male and female flowers of intramorphs, GA-related genes may regulate the asynchronization by perceiving the position differences of two sexual buds. As stated by Mao et al. [12], in C. paliurus female buds grow at the apex of the growing shoot and male buds locate at the lateral short branch; the expression of GA-related DEGs in male and female buds may be triggered under the genetic control depending on the mating type.  (Figure 4), which is consistent with the phenomenon that the physiological differentiation of female buds was not complete until S2 in PA. However, DELLA was significantly upregulated in the comparison of PA-M vs. PA-F at S2, confirming the functions of DELLA proteins on inhibiting vegetative and reproductive growth [40,41]. For the PG-F vs. PG-M comparison, GID1 showed a higher expression level and DELLA showed a lower level in PG-M at S2, which indirectly supported that GA was more necessary for male buds germination than female.
Previous studies have proved that gibberellins have been implicated in coordinating flower differentiation and development by regulating various cellular changes in response to environmental or intrinsic signals. For instance, among GAs, GA 1 , GA 4 , and GA 7 were reported to partially inhibit floral initiation in Prunus avium [42], while bioactive GA 4 was necessary for female flower development in Cucumis [43]; moreover, GA 3 not only slightly delayed senescence but also affected flower opening in Iris × hollandica cv. Blue Magic [44]. Furthermore, exogenous GA 3 application partially restored the inhibitory effects of heavy metals on flowering yield in Lagenaria siceraria (Mol.) Standl and Luffa cylindrica L. (Roem) [45], and significantly delayed floral induction in 'Fuji' apples (Malus domestica Borkh.) [46]. In this study, in contrast to other GAs in four types of floral buds (PG-F, PG-M, PA-F, PA-M), GA 3 content showed significantly higher values with apparent variations over five stages (Table 1), suggesting its critical role in controlling flowering bud differentiation and development. At the S0 stage, a low level of GA 3 in female buds and a high level in male buds in PA ( Figure 5) is in accordance with the anatomical observations (data unpublished), which revealed the differentiated male buds and undifferentiated female buds. On the contrary, the high GA 3 content of both floral buds in PG ( Figure 5) was also consistent with the anatomical findings, which showed differentiated floral buds but female before male buds.
In addition, higher GA 3 contents in male over female buds in both the PA and PG types at two stages infers that the activity of male buds could be more dependent on high GA 3 . Collectively, these results from C. paliurus are consistent with previous studies (the rhythm in Kalanchoe spp., Brassica rapa ssp. chinensis Makino), that GA 3 promoted floral induction and germination and even affected the number of days taken to flower [47][48][49]. Quite the reverse, the critical role of GA 3 in controlling the physiological differentiation in male buds and germination in C. paliurus was also verified in Arabidopsis thaliana, whose stamen development required higher GA contents than other floral organs (pistil, petal, and calyx) [50]. Moreover, a higher level of GA 3 in male floral buds at stage 8 (mature pollen) in Diospyros kaki Thunb. was considered to promote the development of stamen primordia [51]. Combining GA 3 content with GA-related DEGs expressions, we deduced that gibberellin response genes potentially play crucial roles in adjusting the asynchronization of the male and female flowers of intramorphs in C. paliurus.

Plant Material and Growth Conditions
An experimental plantation of C. paliurus was established in Hongya Mountain Forest Farm, Chuzhou, Anhui Province, China (N 31 • 21 , E 118 • 58 ) in 2013, containing 800 seedlings from a germplasm bank, which were planted in the Baima Experimental Base of Nanjing Forestry University. Voucher specimens were identified by Xin Chen (a taxonomist of Nanjing Forestry University) and deposited in the Silviculture Lab of Nanjing Forestry University (voucher code: 2011GX, 2011SC, 2011HB, 2011ZJ, 2011HN). Nine protogynous (PG) and protandrous (PA) individuals, respectively, were randomly selected and marked for phenological observation and sampling.
Leaves and floral buds in various protandry (PA) and protogyny (PG) individuals were collected at five different stages of floral differentiation and development including (1) S0, the physiological differentiation period (26 May 2019); (2) S1, the dormancy period (23 December 2019); (3) S2, the germination period (15 March 2020); (4) S3, the inflorescence elongation period (18 April 2020); and (5) S4, the maturation period (28 April 2020). Among them, only male and female floral buds from two mating types were collected as experiment materials at S1, S2, and S3, because of their deciduous characteristic. However, during the other two phases, floral buds and leaves were harvested for GAs content determination. Leaves at S4 and floral buds at five stages were collected for transcriptome sequencing. All flower buds were invisible during physiological differentiation (S0), so the flower buds were harvested based on last year's position on the branch and the duration of morphological observation was from the dormancy stage (S1) to the maturation period (S4). Samples of complementary gender in each mating type were harvested from three individuals and combined as one independent biological replicate, and the total of nine trees were grouped as three replicates. Collectively, more than 15 male flower buds and at least 45 female flower buds were collected for each replicate, respectively. All samples were preserved in liquid nitrogen and stored at −80 • C for further analysis.

RNA Extraction and Illumina Sequencing
Total RNA was extracted from various tissues using an E.Z.N.A Plant RNA Isolation Kit (OMEGA, Norcross, GA, USA). RNA concentration and purity were determined by a NanoDrop spectrophotometer 2000C (IMPLEN, Westlake Village, CA, USA), and RNA integrity was confirmed by 1.0% agarose gel electrophoresis. The RNAs with the values of OD 260 /OD 280 within 1.8-2.0 were reverse transcribed to cDNA using a PrimeScript™ RT reagent kit with gDNA Eraser (TaKaRa, Beijing, China), according to the instructions of the manufacturer. Then, the cDNAs were diluted at 1:10 with RNase-free water, and stored at −20 • C for qRT-PCR analyses. All cDNA libraries were loaded onto an Illumina HiSeqTM2000 system (2 × 100 bp read length). Finally, 6 Gb of RNA-seq raw data for each sample was generated, and the quality of raw reads was checked by FastQC [52]. Clean reads were obtained from removing poor-quality sequences (the bases of Q-score < 10%), short sequences (<50 bp), primer sequences, unidentified nucleotides (more than 10%), and adapter sequences.

Assembly and Differentially Expressed Genes (DEGs) Analysis
Paired-end short reads were aligned to the PG-dip C. paliurus genome using Hisat2 [53]. StringTie (v.1.3.3b) [54] was used to assemble transcripts and estimate the gene expression levels of each sample. The expected number of FPKM (fragments per kilobase per million reads) fragments mapped were calculated using the RSEM program [55], which was implanted in the Trinity package [56]. Moreover, the DESeq2 R package [57] was applied to identify the DEGs. In multiple analyses, the threshold of the p-value was calculated by the false discovery rate (FDR). If the FDR was less than 0.05, and the absolute value of log 2 (fold change) was more than 1, the gene expression differences were considered to be significant.
The Gene Ontology (GO) functional annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed using the clusterProfiler R package [58].

Measurement of GAs Content
A UPLC-MS/MS system (UPLC, SHIMADZU Nexera X2; MS/MS, Applied Biosystems 4500 Q TRAP) was used to determine the content of four types of GAs (bioactive GA 1 , GA 3 , GA 4, and GA 7 ) at different developmental stages.
Standards (gibberellin, GA 3 ) were purchased from Sigma-Aldrich (St. Louis, MO, USA), while MeOH and ACN were purchased from Merck (Darmstadt, Germany). C. paliurus vacuum freeze-dried floral buds and leaves were ground into powder (30 Hz, 5 min), and 500 mg of powder was extracted with 1.5 mL of methanol/water/methanoic acid solution (15:4:1, V/V/V). The extract was vortexed every 30 min and stored at 4 • C for one night. After centrifugation (rotating speed 12,000 rpm, 10 min), the supernatants were filtered with a microporous membrane (0.22 µm pore size) and collected for UPLC-MS/MS analysis.
The UPLC operation parameters were as follows: chromatographic separation was carried out using a Waters ACQUITY UPLC HSS T3 (100 mm × 2.1 mm, 1.7 µm) and the mobile phase consisted of ultrapure water (A) and acetonitrile (B), which both contained 0.1% formic acid. Gradient elution was as follows: the original B ratio was 5%; the B ratio linearly increased to 70% within 3 min and was maintained for 3 min; and the B ratio decreased to 5% from 6.1 to 8 min and was maintained for 1.9 min. The flow rate was kept at 0.2 mL/min, temperature at 40 • C, and injection volume was 2 µL.

Quantitative Real Time PCR (qRT-PCR) Analysis
Eight unigenes related to floral bud differentiation and development, as well as the hormone-mediated signaling pathway, were selected for qRT-PCR analysis. Moreover, 18S ribosomal RNA as an internal control and the cDNA Synthesis Kit for reverse transcription were selected according to Chen [59]. Proprietary software with Quant Studio 6 (Life Technologies, Carlsbad, CA, USA) and GoTaq ® qPCR Master Mix (PROMEGA, Cat.No.A6002) were used for RT-qPCR reaction. The reaction system was as follows: 2× MasterMix 10 µL, Primer F 1 µL(10 µM), Primer R 1 µL(10 µM), cDNA 1 µL, and nuclease-free water up to 20 µL. The reaction procedure and methods of relative gene expression calculation were presented by Chen [59].

Conclusions
This study reveals that hormones, especially GA 3 , may play an important role in contributing to the heterodichogamy morphs with asynchronous flowering in C. paliurus. The DEGs during five stages were significantly enriched in a series of biological processes involved in flower and organ development, hormone biosynthetic and metabolic processes, and GA signaling pathways. GA 3 content showed a significant role in controlling flowering bud differentiation and development. Our results can provide the basis for further understanding the molecular mechanisms for gibberellin regulating floral bud differentiation, development, and their asynchronism in heterodichogamous plants. Differentially expressed genes qRT-PCR Quantitative real-time polymerase chain reaction FDR False discovery rate FPKM Fragments per kilobase per million reads