Morphology analysis of petal color transitions
Petal color of every single flower was transformed continuously from green to white and then to yellow during flower development in L. japonica. Early in the development of floral buds, primary bud with green petal grew into approximate 3.5-cm in length (Fig. 1a). At the early stage of anthesis, the petals turned from green to white (Fig. 1b). Then, the petals gradually transformed into yellow from white before the withering stage (Fig. 1c). During petal color transitions, the petals at the stages of green bud, white flower and yellow flower were selected. Changes of color index of GB_Pe, WF_Pe and YF_Pe were significant differences (Table S2). Value of redness (a*) in GB_Pe, WF_Pe and YF_Pe were -12.36, -0.58 and 1.25, respectively. Parameter lightness (L*) in WF_Pe was 82.40, which was higher than that in GB_Pe and YF_Pe. Index of yellowness (b*) in the YF_Pe was the highest (42.55).
Carotenoid accumulations in L. japonica petals at various stages
To obtain an accurate understanding of the carotenoid accumulation, carotenoid profiling was analyzed in L. japonica petals using LC-MS/MS during petal color transitions. A total of 13 carotenoids were detected from GB_Pe, WF_Pe and YF_Pe (Table 1). Most of carotenoids were significantly highly accumulated in YF_Pe. Compared with GB_Pe and WF_Pe, the α-carotene, antheraxanthin, lycopene, zeaxanthin, violaxanthin, γ-carotene, neoxanthin, β-carotene, β-cryptoxanthin and apocarotenal were significantly up-regulated in YF_Pe. Among them, violaxanthin was the major carotenoid compound in YF_Pe. Some carotenoid compounds, such as lycopene, γ-carotene, β-carotene and β-cryptoxanthin were only detected in YF_Pe.
Widely targeted flavonoid profiling of L. japonica petals at different stages
To better understand the content changes of flavonoid, quantitative analysis of flavonoids was further performed by LC-MS/MS technology. A total of 41, 43 and 37 significantly differentially accumulated flavonoids were identified in WF_Pe vs GB_Pe, YF_Pe vs GB_Pe and YF_Pe vs WF_Pe comparisons, respectively (Fig. 2a and Table S3). In total, ten anthocyanins were identified in all samples. Among them, pelargonidin, cyanidin, cyanidin O-malonyl-malonylhexoside and delphin chloride were significantly differentially accumulated. Compared with GB_Pe and WF_Pe, pelargonidin and cyanidin was significantly up-regulated in YF_Pe. Specifically, pelargonidin was not detected in GB_Pe. Compared with WF_Pe, the contents of pelargonidin and cyanidin were increased by 2.11- and 2.36-fold in YF_Pe, respectively. However, cyanidin O-malonyl-malonylhexoside and delphin chloride were not detected in YF_Pe (Fig. 2b and Table S3).
Effects of endogenous hormones during petal color transitions
To obtain the changes of endogenous hormones, the concentrations of indoleacetic acid (IAA), zeatin riboside (ZR), GA, BR, methyl jasmonate (MeJA) and ABA were analyzed. During petal color transitions, the concentrations of IAA, ZR, GA, BR and MeJA decreased, but the content of ABA increased (Fig. 3). The IAA concentration decreased significantly from 717.3 ng·g-1 FW (GB_Pe) to 191.0 ng·g-1 FW (WF_Pe) then to 118.8 ng·g-1 FW (YF_Pe). The ZR and GA concentrations were both firstly decreased significantly from GB_Pe to WF_Pe, and then keep stable from WF_Pe to YF_Pe. The BR concentration was highest in GB_Pe. From GB_Pe to YF_Pe, the BR concentration decreased significantly from 9.2 ng·g-1 FW in GB_Pe to 7.3 ng·g-1 FW in WF_Pe, and then increased slightly to 8.3 ng·g-1 FW in YF_Pe. The level of MeJA was firstly decreased significantly from GB_Pe to WF_Pe reaching the lowest level, and then slightly increased from WF_Pe to YF_Pe. However, the ABA concentration increased significantly from 98.0 ng·g-1 FW to 205.2 ng·g-1 FW from GB_Pe to YF_Pe (Fig. 3).
Sequencing, de novo assembly and annotation
To identify key candidate genes for petal color transitions, RNA sequencing was carried out from GB_Pe, WF_Pe and YF_Pe. Nine cDNA libraries were sequenced and 448 565 884 raw reads were generated. After the data filtering, 408 576 816 (91.1%) clean reads were produced and Q30 values were greater than 96.7%. For each sample, clean reads were obtained from 6.6 to 7.1 Gb (Table S4). A total of 69 946 unigenes were generated with average length is 871 bp and an N50 is 1 636 bp (Table S5). Most of unigenes (96.6 %) were generated from 200 to 3 200 bp in length, and 2 383 (3.4%) unigenes were more than 3 200 bp (Fig. S1).
A total of 34 068 assembled unigenes were annotated (Table S6). Based on sequence similarity, 22 662 (32.4 %) unigenes were enriched into three groups (biological process, cellular component and molecular function) based on GO term analysis (Fig. S2). The biological processes were mainly focused on ‘cellular process’ and ‘metabolic process’. The cellular components were mainly involved in ‘cell part’. The molecular functions were mainly classified into ‘binding’ and ‘catalytic activity’. KEGG term analysis was used to identify the functional classifications of the unigenes. There were 9 309 (13.31%) unigenes were enriched into 32 KEGG pathway groups, of which ‘signal transduction’ represented the largest group, followed by ‘carbohydrate metabolism’, ‘translation’ and ‘folding, sorting and degradation’ (Fig. S3).
Identification and analysis of DEGs
To detect alterations in gene expression, transcriptomic analyses of WF_Pe vs GB_Pe, YF_Pe vs WF_Pe and YF_Pe vs GB_Pe were carried out to identify the key DEGs during petal color transition in L. japonica (Fig. S4). A total of 29 679 DEGs were identified based on a 2-fold change at P < 0.05 (Fig. S4a). The comparison of WF_Pe vs GB_Pe showed a total of 22 932 DEGs were identified, of which 10 013 were up-regulated and 12 919 were down-regulated. In the YF_Pe vs GB_Pe comparison, 18 984 DEGs were identified, of which 8381 were up-regulated and 10 603 were down-regulated. The comparisons of YF_Pe vs WF_Pe showed a total of 12 220 DEGs were identified, of which 5 936 were up-regulated and 6 284 were down-regulated (Fig. S4b).
All identified 29 679 DEGs were further classified into 8 clusters on the basis of expression alteration during petal color transition (Fig. S5a). A total of 3 470 DEGs were classified into two profiles based on expression changes across the three development stages: expression stable and then increased (profile 4) and expression stable and then decreased (profile 3). The opposite change patterns of gene expression during petal color transition from white to yellow, suggesting a tight linkage of these genes with petal color transition in L. japonica.
GO enrichment analysis was further performed to investigate biological functions of these 1 897 DEGs (RPKM > 1 in at least one sample from the 3 470 DEGs) that showed higher or lower expression in YF_Pe. The hormone-mediated signaling pathway was significantly enriched in biological process subcategory (Fig. S5b). DEGs involved in hormone-mediated signaling pathway, including small auxin-up RNA (SAUR) and PYRABACTIN RESISTANCE1-like (PYL), were significantly differentially expressed between yellow petals and non-yellow petals and seemed the most relevant to the goal of this study.
Enrichment analysis of DEGs involved in hormone-mediated signaling pathway
GO enrichment analysis showed that DEGs were mainly enriched in hormone-mediated signaling pathway. To better investigate hormonal regulation in the color transitions, we analyzed the 67 DEGs (>1 RPKM) that were enriched in the signaling pathways of auxin, cytokinin, gibberellin (GA), brassinosteroid (BR), jasmonic acid (JA), ABA and ethylene in YF_Pe vs GB_Pe and YF_Pe vs WF_Pe (Fig. S6 and Table S7).
In the auxin signaling pathway, 15 DEGs were identified, of which AUX1, TIR1, ARF and SAUR genes were significantly downregulated from GB_Pe to YF_Pe, while three IAAs were upregulated at WF_Pe (Fig. S6a). A total of 18 DEGs was enriched in the cytokinine signaling pathway, including HKs, HPs, type-B RRs and type-A RRs. All of these DEGs were downregulated from GB_Pe to WF_Pe and YF_Pe (Fig. S6b). Meanwhile, in the GA signaling pathway, GID1, GID2 and DELLA genes were identified and significantly downregulated in YF_Pe (Fig. S6c). In the BR signaling pathway, 13 DEGs were identified, most of them were firstly downregulated and then upregulated in the transition. Specifically, the expression of BRI1, BSK, BZR1_2, CYCD3 and TCH4 were significantly higher in YF_Pe than in WF_Pe (Fig. S6d). Four DEGs were enriched in the JA signaling pathway, and their expression levels were higher in GB_Pe than in WF_Pe and YF_Pe (Fig. S6e). Furthermore, JAR1, COI-1 and MYC2 were expressed at higher levels in YF_Pe than in WF_Pe, while JAZ was expressed at lower levels in YF_Pe than in WF_Pe. However, seven DEGs were identified in the ABA signaling pathway, including PYL, PP2C, SNRK2 and ABF, of which PYLs and SNRK2 were significantly upregulated in YF_Pe (Fig. S6f). In the ethylene signaling pathway, five DEGs were identified, of which EIN3, ERS and ERF1 were significantly upregulated in YF_Pe (Fig. S6g).
Pigments accumulation regulation ofpetal color transitions
To investigate the pathways of pigments synthesis/degradation during the transitions, expression levels of carotenoid, flavonoid and chlorophyll metabolism-related genes were analyzed. A total of 44 DEGs (>1 RPKM) regulating carotenoid, flavonoid and chlorophyll metabolism were identified and significantly differentially expressed between yellow petals and non-yellow petals (GB_Pe or WF_Pe) (Fig. 4 and Table S7).
In the carotenoid biosynthesis pathway, PSY, PDS and ZDS were significantly upregulated in YF_Pe. However, three carotenoid degradation-related genes, including carotenoid cleavage dioxygenase 4 (CCD4), CCD7 and abscisic-aldehyde oxidase 3 (AAO3), were significantly downregulated in YF_Pe (Fig. 4a). Meanwhile, expression levels of chlorophyll metabolism-related genes showed significant difference. Among these genes, biosynthesis-related genes, including glutamyl-tRNA synthetase (GltX), protoporphyrinogen IX oxidase (PPO) and chlorophyll synthase (CHLG) were significantly upregulated in GB_Pe. However, pheophytinase (PPH), pheophorbide a oxygenase (PAO) and red chlorophyll catabolite reductasewere (RCCR) were significantly downregulated in GB_Pe (Fig. 4b). In addition, expression levels of flavonoid/anthocyanin biosynthesis-related genes, such as chalcone synthase 2 (CHS2), flavonoid 3'-monooxygenase (F3’H), anthocyanidin 3-O-glucoside 5-O-glucosyltransferase (UGT75C1) and dihydroflavonol 4-reductase (DFR), were significantly downregulated in YF_Pe. Meanwhile, caffeoyl-CoA O-methyltransferase (CAMT) was expressed at lower level in YF_Pe than in GB_Pe (Fig. 4c).
Validation of the expression analysis of key pigment-related genes
A total of ten pigment-related unigenes were randomly selected and identified by RT-qPCR. The expression patterns of these DEGs were corresponded well with the RPKM values obtained by RNAseq (Fig. 5). Pearson correlation analysis showed high correlation coefficients between RNA-seq and RT-qPCR data, suggesting the sequencing data are reliable.