Time-Series Transcriptome Sheds Light on Anthocyanin Metabolism and Species Diversication Between two Lonicera Japonica Thunb Cultivars

Background (cid:0) ‘Yujin 2’ is new variety of Lonicera japonica Thunb and its ower color can change from red to yellow; hence, it is a good model for investigating ower color development mechanisms. Results (cid:0) High throughput transcriptome sequencing of seven ower development stages of Yujin No.2 was carried out, and 133,487 unigenes were annotated, among which 73,088 were differentially expressed. Then the real-time PCR analysis was carried out. Further, the number of up-regulated DEGs was higher than those that were down-regulated. Of these annotated DEGs, plant hormone signal transduction, phenylpropanoid biosynthesis, and avonoid biosynthesis were active throughout the owering process during each stage, whereas carotenoid biosynthesis was inactive in the S1-6 stages. Furthermore, phenylalanine synthesis was enhanced in the S1 phase; however, anthocyanin synthesis was weakened in the S5 and S6 phases, which may be consistent with the changes in petal color of ‘Yujin 2’ from red (S1) to white (S5) and gold (S6). The results showed that 114 unigenes were associated with anthocyanin metabolism, and 72 were signicantly upregulated or downregulated. According to the analysis of TFs in anthocyanin metabolism, we obtained 47 transcription factors, which belonged to 18 families. The LjDFR, LjABCB1, LjMYC6, LjDDB2, and LjANS genes rapidly increased during the rst three stages. However, only LjF3'5'H expression was signicantly down-regulated at S5, which was consistent with anthocyanin accumulation. germplasm


Background
Lonicera japonica Thunb (L. japonica) is a perennial semi-evergreen twining stoloniferous shrub. Traditionally, the L. japonica ower bud, which has been listed in the Chinese Pharmacopoeia as Lonicera japonica (Jinyinhua in Chinese) is considered to be the primary medicinal component [1]. Recent studies have found that L. japonica has many medicinal properties due to its antioxidant [2], hypoglycemic and hypolipidemic [3], anti-allergic [4], anti-in ammatory [5], and antibacterial effects [6]. Further, the demand for L. japonica in pharmaceuticals [1], cosmetics [7], and health foods [8] has increased. Although L. japonica is widely distributed in China, its yields and quality are the highest in Fengqiu of Henan and Shandong Provinces. Therefore, our research group selected and bred a new variety (Yujin 2). It has the characteristics of red buds, strong resistance, and a high content of chlorogenic acid (CGA) and luteoloside. The red buds endow 'Yujin 2' with enhanced beauty, which is more valuable in utility [9].
As one of the most important phenotypic traits of ornamental plants, the formation of ower color is the result of the combined effect of environmental in uences and genetic characteristics [10]. The modi cation of petal color regulation genes is indispensable in modern ower breeding [11]. Through various studies it has been established that changes in ower color are primarily due to alterations in plant pigment metabolism [12]. The key pigments to adjust the color are carotenoids, avonoids, and alkaloids [11; 13; 14].
Anthocyanins, which are produced through avonoid biosynthesis pathways during plant metabolism, are important water-soluble natural food pigment, which is widely present in plant epidermal cell vacuoles [15]. In the rst stage of anthocyanin biosynthesis, phenylalanine transforms into 4-coumaryl:CoA, referred to as the phenylpropanoid metabolic pathway. The next stage, 4-coumaryl:CoA transforms into multiple avonoid compounds [16].
The biosynthesis of anthocyanins is related to a variety of structures and regulatory genes [17]. The expression of chalcone synthase (CHS), Flavanone 3-hydroxylase (F3H), dihydro avonol-4-reductase (DFR), anthocyanin synthase (ANS) and UDP gluco avonoid 3-o-glucosyltransferase (UFGT) directly affect the accumulation of anthocyanins in plants [18]. These structural genes have been cloned in different plants, and the expression pathway is common to different types of plants.
Key regulatory genes serve as regulators that combine individually, or in the form of dimers and trimers, which play a critical role in the regulation of anthocyanin synthesis. So far, MYB, bHLH and WD40 transcription factors are widely regarded as the three main regulatory factors [19]. The MYB family transcription factor is one of the largest transcription factor families in plants, which has a MYB domain that is composed of 1 ~ 4 MYB repeating units in its structure [20]. In conjunction with MYB, the bHLH transcription factors play active and critical roles in anthocyanin biosynthesis. The high anthocyanin content of many crops results from the increased expression of bHLH transcription factors [10]. WD40 is an ancient and highly conserved family of proteins that is involved in myriad plant cellular processes, such as transcriptional regulation, cell division, vesicle formation and transport, processing, signal transduction and so on [21].
There were great differences in the corolla color of 'Yujin 2' at the various developmental stages. The upper half of the ower bud during the early stage was purple-red, whereas the whole ower bud from the three green periods to the great white period was red, but gradually became lighter; whereas during the silver period, the exterior was red and the interior was white. During the golden period, the exterior of 'Yujin 2' was red and the interior was white, but for 'Damaohua' it was yellow [9].
Color is considered to be an important indicator in the assessment of ornamental value and offers a basis for plant classi cation [11]. However, the molecular mechanisms of anthocyanin metabolism in L. japonica owers have not yet been discovered. Therefore, this paper employed 'Yujin 2 and 'Damaohua' owers at different stages of development to study the molecular mechanisms of anthocyanin formation. This, by means of morphology, transcriptome, and bioinformatics, etc., to provide possibilities for enhancing the L. japonica germplasm using genetic engineering technologies, and cultivating new varieties with different colors.

Characteristics of 'Damaohua' and 'Yujin 2' transcriptome
A total of 42 samples ('Damaohua' and 'Yujin 2', respectively, have seven stages, each with three independent replicates) were sequenced using the Illumina HiSeq X Ten sequencer (Illumina Inc., USA), from which 311 Gb of raw data was harvested. Using a pair-end sequencing strategy, 2G raw reads with mean lengths of 150 nt were obtained from the 42 samples with 47.62 M reads for each sample. After the ltering of raw reads, 1.95 G reads were counted, with an average of 46.43 M reads per sample (Additional le 1: Table S1). Following the removal of redundant sequences and quantity control, 133,487 unigenes were assembled. The average length of the unigenes was 1,047 bp with 45.75% GC and 1642nt N50. The longest transcript was 16,864 bp, and the shortest was 301 bp.
The unigenes were annotated to obtain functional information using seven separate databases, including NCBI non-redundant protein (NR), Clusters of Orthologous Groups (COG) for eukaryotic complete genomes, Gene Ontology (GO), Swiss-Prot, evolutionary genealogy of genes: Non-supervised Orthologous Groups (eggNOG), and the Kyoto Encyclopedia of Genes and Genomes (KEGG) databases by Diamond Software, and mapped to Pfam databases by HMMER. Overall, 66.79% (133,487 unigenes) were successfully annotated, among which most unigenes (66.46%) were assigned to annotation terms in the NR database (Table 1) According to the NR annotation results, the current study compared L. japonica with taxonomic species stored in the NR database. A total of 97247 sequences were matched with L. japonica unigenes, which covered 1512 taxonomy categories (Additional le 2: Table S2). The highest ranked species was Alternaria alternate, which took up 12.82%, whereas the second highest-ranked was Vitis vinifera, which contributed 8.80% (Fig. 1).

Differentially expressed genes (DEGs)
When comparing seven owering stages of expressed genes between 'Damaohua' and 'Yujin 2', 73,088 differentially expressed genes exhibited remarkable changes. A total of 5,355 genes were differentially expressed, with 2,864 genes up-regulated and 2491 genes down-regulated at the S1 stage. At the S2 KEGG-pathway analysis of differentially expressed genes Unigenes were assigned to KEGG pathways based on annotation data. DEGs were enriched into 181, 202, 202, 191, 207, 204, and 210 pathways (S1, S2, S3, S4, S5, S6, and S7) (Additional le 3: Table S3). At the S1 stage, 53 pathways were enriched with a statistically signi cant enrichment p-value. Plant hormone signal transduction (ko04075), and phenylpropanoid biosynthesis (ko00940) during the entire process, as well as up-regulation and down-regulation processes, played active roles, and the expression was signi cant. Amino sugar and nucleotide sugar metabolism (ko00520) were also active during the entire process and up-regulation, while they were not active in the down-regulation process. Starch and sucrose metabolism (ko00500) were active during the entire process and down-regulation process, while they were not active in the up-regulation process (Figs. 1A).
At the S2 stage, 72 pathways were enriched with a statistically signi cant enrichment p-value, which was more than the other six paired comparison phases. Endocytosis (ko04144), plant hormone signal transduction (ko04075), and starch and sucrose metabolism (ko00500) during the entire process, as well as up-and down-regulation processes played an active role, and the expression was signi cant. Phenylpropanoid biosynthesis (ko00940) was active during the entire process and the down-regulation process, while it decreased in the up-regulation process (Figs. 1B).
At the S3 stage, 66 pathways were enriched with a statistically signi cant enrichment p-value. Plant hormone signal transduction (ko04075), phenylpropanoid biosynthesis (ko00940), starch and sucrose metabolism (ko00500) during the entire process, and the up-and down-regulation processes played an active role, and the expression is signi cant. Glycolysis/Gluconeogenesis (ko00010) was active during the entire process and the down-regulation process, while it was not active in the up-regulation process (Figs. 1C).
At the S4 stage, 71 pathways were enriched with a statistically signi cant enrichment p-value, and plant hormone signal transduction (ko04075) during the entire process, the up-and down-regulation processes played an active role, and the expression was signi cant. Phenylpropanoid biosynthesis (ko00940) and cell cycle (ko04110) were active during the entire process and the up-regulation process, while they were not active in the down-regulation process. Starch and sucrose metabolism (ko00500) were active during the entire process and the down-regulation process, while they were not active in the up-regulation process. Cell cycle-yeast (ko04111) was triggered in the up-regulation process, while carbon metabolism (ko01200) was triggered in the down-regulation process (Figs. 1D).
At the S5 stage, 63 pathways were enriched with a statistically signi cant enrichment p-value. Plant hormone signal transduction (ko04075) and phenylpropanoid biosynthesis (ko00940) played active role during the entire process as well as up-and down-regulation processes, and the expression was signi cant. Glycerophospholipid metabolism (ko00564), starch and sucrose metabolism (ko00500) were active in the total process and the up-regulation process, which were not active in the down-regulation process. Glutathione metabolism (ko00480), drug metabolism -cytochrome P450 (ko00982), and metabolism of xenobiotics by cytochrome P450 (ko00980) were triggered in the down -regulation process (Figs. 1E).
At the S6 stage, 31 pathways were enriched a with statistically signi cant enrichment p-value. Plant hormone signal transduction (ko04075), phenylpropanoid biosynthesis (ko00940), starch and sucrose metabolism (ko00500), glycolysis/gluconeogenesis (ko00010) played an active role during the entire process, the up-and down-regulation process, and the expression was signi cant. Valine, leucine and isoleucine degradation (ko00280) was triggered in the up-regulation process (Figs. 1F).
At the S7 stage, 39 pathways were enriched with a statistically signi cant enrichment p-value. Fatty acid metabolism (ko01212) was active during the entire process, but was not in the up-and down-regulation process. Plant hormone signal transduction (ko04075), phenylpropanoid biosynthesis (ko00940), starch and sucrose metabolism (ko00500) were active during the entire process and up-regulation process, while they were not in the down-regulation process. Tyrosine metabolism (ko00350) was active during the entire process and the down-regulation process, while it was not in the up-regulation process. Spliceosome (ko003040) was triggered in the up-regulation process, whereas MAPK signaling pathwayyeast (ko04011), as well as glycine, serine and threonine metabolism (ko00260) were triggered in the down -regulation process (Figs. 1G).
Summarizing the above KEGG pathways enriched during the owering period, it was found that phenylpropanoid biosynthesis and avonoid biosynthesis were active through the entire owering process and nearly up-regulated at each stage, Peroxisome(ko04146), Carbon metabolism ko01200 and Amino sugar and nucleotide sugar metabolism ko00520 can be found in each period (Fig. 3A) ,and carotenoid biosynthesis was down-regulated from S1-6. The anthocyanin precursor, phenylalanine, was enhanced at the S1 phase; however, the synthesis of anthocyanin was weakened at the S5 and S6 phases ( Table 2). Gene Ontology (GO) analysis of differentially expressed genes Differentially expressed genes were annotated by Gene Ontology terms (Additional le 4: Table S4). Genes were classi ed into three GO categories, that is, Biological Process, Cellular Component, and were found at the every period (Fig. 3B) Anthocyanin metabolism analysis of differentially expressed genes Sequencing results revealed that 114 unigenes were associated with anthocyanin metabolism, and 72 were signi cantly upregulated or downregulated across the seven owering stages. The number of upregulated unigenes was highest at the S7 stage, while the number of down-regulated unigenes was highest at the S5 stage (Fig. 4A). According to the analysis of TFs in anthocyanin metabolism, we obtained 47 genes relating to transcription factors in anthocyanin metabolism that belonged to 18 families, of which bHLH had the largest number, followed by GeBP (Additional le 5: Table S5, Fig. 4B). According to the anthocyanin heatmap analysis, the expression of genes at S7 was different from that at other stages, with the highest up-regulated expression. The expression of genes at S6 was similar to that at S5, while those at S4, S3, S2, and S1 were similar, among which the expression of genes in S1 and S2 were the most similar (Fig. 4C).

qRT-PCR veri cation
To validate the accuracy and repeatability of our RNA-Seq data, the current study selected six genes for quantitative real-time polymerase chain reaction (qRT-PCR) analysis, with gene-speci c primers designed using Primer software (version 5.0), as shown in Additional le 6: Table S6. As shown in Fig. 5, the results of qRT-PCR indicated that most of these genes had expression patterns that correlated with the RNA-Seq data, which con rmed the reliability of our data. The genes LjDFR (TRINITY_DN37612_c0_g1_i2_29), LjABCB1 (TRINITY_DN23592_c0_g2_i2_8), LjMYC6 (TRINITY_DN17985_c0_g1_i2_8), LjDDB2 (TRINITY_DN25019_c0_g2_i6_8), and LjANS (TRINITY_DN27734_c0_g1_i1_12) rapidly increased during the rst three stages. However, the expression of only LjF3'5'H (TRINITY_DN27086_c0_g1_i1_29) was signi cantly down-regulated at S5, which was consistent with anthocyanin accumulation.

Discussion
The biological regulation of ower color had garnered much attention from researchers, particularly in terms of ornamental plant research [11]. This present study focused on a new variety of L. japonica (Yujin no. 2). There are a variety of designs and colors for 'Yujin 2' that has a high ornamental value, which may play a signi cant role in landscaping [9]. It is very important to regulate anthocyanin synthesis in L. japonica breeding [22]. With respect to molecular breeding, the identi cation of the key genes involved in anthocyanin synthesis will enable the overexpression or knockout of these genes to facilitate color modi cation [23]. Therefore, this study began with the aim of identifying the essential genes that regulated the alteration of ower color and subsequently exploring the potential molecular mechanisms behind the development of ower color.
However, little evidence has been reported as relates to L. japonica petal color. The current study results provide a foundation for research into petal color mechanisms at the transcriptome level. Transcriptome analysis based on an EST sequencing dataset illustrated the characteristics of the carnation transcriptome and revealed that carotenoid, chlorophyll, and anthocyanin biosynthesis played roles in the regulation of carnation ower color [24].
Correlation analysis showed that the changes in Paeonia ower color, from coral to pink to pale yellow were due to a signi cant decrease in anthocyanin content [25]. It was deduced that the disequilibrium of expression levels in avonol synthases and dihydro avonol-4-reductases resulted in different levels of anthocyanin accumulation and coloration in white and pink tea owers via metabolome and transcriptome analysis [26]. Besides, the transcriptome dataset assisted with the establishment of linkages between gene expression and other aspects, including the phenotypical, biochemical and metabolic features of Meconopsis petal development [10].
Petal color change is a signi cant feature of gene expression. As a whole, the number of differentially expressed genes at each owering development stage was variable, which revealed that transcriptome expression changes were more active during color alteration. For example, when the petal color changed from purple-red (S1) to red (S2), 15804 genes were differentially expressed; while only 2033 genes changed from S2 to S3, when the petal color at both were red. Further, when the petal colors were red, but gradually became lighter from the three-green period (S2) to the great white period (S4), the number of differentially expressed genes were eventually lower, which were 21159, 19126, and 11153, respectively. Besides, gene expression changes occurred when the petal color transitioned from red (S4), to the outer part being red and the inner part being white (S5), where 6073 genes were up-regulated from the bud stage to cracking stage (Fig. 2).
Petal color phenotypes result from plant pigment metabolism [13]. Anthocyanin, avonoid, and alkaloid are well-known plant pigments that play active roles in the regulation of ower color [11]. Flavonoids are involved in the production of red, pink, purple and blue in plants and are widely distributed in petals [27].
Additionally, and most importantly, avonoid accumulation in a spontaneous cotton mutant resulted in red coloration and enhanced disease resistance [28]. Anthocyanins are a group of water-soluble pigments that confer the blue, purple and red color to many fruits [29]. Carotenoids primarily affect yellow, orange and red coloration [30]. To identify the potential mechanisms that regulate the modi cation of owering colors, the current study annotated differentially expressed genes via KEGG pathway analysis and Gene Ontology.
In the pathway analysis, several pigment metabolism pathways were highlighted, including plant hormone signal transduction (ko04075), phenylpropanoid biosynthesis (ko00940), anthocyanin biosynthesis (ko00942), avonoid biosynthesis (ko00941), and carotenoid biosynthesis (ko00906). Furthermore, plant hormone signal transduction, phenylpropanoid biosynthesis, and avonoid biosynthesis were active throughout the owering process and nearly up-regulated at each stage. These results corroborated the ndings of several previous investigations into the linkages between petal color and plant pigment. Previous studies found that the owers of 'Yujin 2' were red due to their higher carotenoid and avonoid content in contrast to 'Damaohua' at all developmental stages [9].
However, the results of our study revealed that carotenoid biosynthesis was down-regulated throughout S1-6, which was different from those of previous studies to some extent, due to the formation of mutants that varied with sampling times. Phenylalanine is an important anthocyanin biosynthesis pathway [13]. In this study, phenylalanine synthesis was enhanced at the S1 phase, and anthocyanin synthesis was weakened at S5 and S6. This result may be consistent with the changes of 'Yujin 2' petal color from red (S1) to white (S5) and gold (S6).
The results obtained in the current study revealed that the phenylpropanoid biosynthetic process (GO:0009699), carotenoid biosynthetic process (GO:0016117), and anthocyanin 5-O-glucoside-4'''-Omalonyltransferase activity (GO:0102801) were triggered. Anthocyanin is the stable con guration of anthocyanidins that is always combined with a sugar moiety [31], which may explain the active glucoserelated metabolism pathways in our research.
GO annotation also supported the KEGG evidence. Furthermore, GO annotation showed that plasma membrane (GO:0005886) was active throughout the owering process. Chloroplast (GO:0009507) and vacuole (GO:0005773) were active through most of the owering processes, and could be explained by the results of previous experiments. The colors of plants are primarily codetermined by the different distributions of chlorophyll, carotenoid, avonoid, and betaine, which exist in the cytoplasms or vacuoles of plants [9]. Moreover, GO analysis revealed that environment-related responses (such as to cold) were triggered (GO:0009409).
Generally, the binding of upstream transcription factors, initiation of gene transcription and expression, nal synthesis of anthocyanins, and coloration of fruits and peels are closely related to the expression of structural genes in the anthocyanin biosynthetic pathway [19]. At present, MYB, BHLH and WD40 are the three major transcription factors currently known. Anthocyanin biological regulation includes singlestructure gene regulation, two-gene interaction regulation, and some MBW (MYB-bHLH-WD40) complexes regulate anthocyanin accumulation [16].
In the rst three stages of ower development, LjDFR, LjABCB1, LjMYC6, LjDDB2, and LjANS genes increased rapidly. However, only the expression of LjF3'5'H was signi cantly down-regulated at S5, which was consistent with the accumulation of anthocyanin. The expression of other transcription factors did not increase during anthocyanin accumulation, which may not be directly related to anthocyanin synthesis. To verify the RNA-Seq results, we randomly selected six genes and tested their expression patterns via qPCR, which revealed the same expression trend with the sequencing results.

Conclusions
In summary, this study developed a transcriptome pro le of ower color generation for L. japonica as well as annotated unigene sets of seven anthesis phases. Bioinformatics analysis revealed a number of distinct clues in regard to color development mechanisms. Further, candidate genes and potential pathways were obtained, which provided a foundation for further research at the transcriptome level. However, the current results still did not fully elucidate the overall regulatory mechanisms of plant pigments and the formation of the red color for L. japonica. Therefore, future studies relating to the current topic, such as transcript regulation, protein expression, and biochemical metabolism, are recommended. Candidate key genes involved in anthocyanin metabolism should also be further investigated.
Our research group collected fresh ower buds of of 'Damaohua' and 'Yujin 2' in seven periods: young bud stage (S1), three green stage (S2), two white stage (S3), great white stage (S4), silver stage (S5), golden stage (S6), fade stage (S7). During sample collection, the owers were combined and regarded as one biological replicate representing each stage, and three independent replicates were prepared. Partial ower materials were immediately frozen in liquid nitrogen following collection and stored at -80 °C.

RNA isolation and library construction
The total RNA was extracted by mirVana miRNA Isolation Kit (Ambion) [32]. RNA integrity was evaluated using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), and samples with the RNA Integrity Number (RIN) ≥ 7 were subjected to subsequent analysis. Separate mRNA from total RNA with oligonucleotide beads (dT) and add fragment buffer to cut it into short fragments, and then use random hexamer primers as templates to synthesize rst strand cDNA. The library was constructed according to the manufacturer's instructions for TruSeq chain mRNA LTSample Prep Kit (Illumina, San Diego, California, USA).

Sequencing, de novo assembly and annotation
The above libraries were sequenced using an Illumina HiSeq X Ten sequencer (Illumina Inc., USA) and 150 bp paired-end reads were generated. The preparation of the cDNA library and sequencing were performed at Shanghai OE Biotech. Co., Ltd., Shanghai, China, and raw data (raw reads) were processed using Trimmomatic [33]. The reads containing ploy-N and low quality reads were removed to obtain clean reads. Following the removal of adaptor and low quality sequences, the clean reads were assembled into expressed sequence tag clusters (contigs) and assembled de novo into transcripts using Trinity [34] (version: trinityrnaseq_r20131110) through a paired-end method. The longest transcript was selected as a unigene based on the similarity and length of a sequence for subsequent analysis.
The function of the unigenes was annotated to the NCBI non-redundant protein (NR), Clusters of Orthologous Groups (COG) for eukaryotic complete genomes, Gene Ontology (GO), Swiss-Prot, and evolutionary genealogy of genes: Non-supervised Orthologous Groups (eggNOG) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases by diamond software, and mapped to Pfam databases by HMMER. The search was conducted using Blastx with a threshold E-value cut-off of 1e-5. Unigene quanti cation, analysis of differentially expressed unigenes (DEGs), cluster analysis, GO and KEGG enrichment FPKM [35], and read count values of each unigene were calculated using bowtie2 [36]and eXpress [37].
We use the DESeq function to estimate the size factor and the nbinom test to determine DEG [38]. The threshold value of P < 0.05 and multiple change > 2 or multiple change < 0.5 was set as the signi cant difference threshold.To explore the transcriptional expression patterns of DEGs, cluster analysis was performed. We performed GO analysis and KEGG pathway enrichment analysis for DEGs based on hypergeometric distribution of R.

qRT-PCR
The same RNA samples used for the RNA-seq experiments were employed for qRT-PCR. We use NanoDrop 2000 spectrophotometer to measure RNA yield (Thermo Scienti c, USA), whereas the integrity was evaluated using agarose gel electrophoresis stained with ethidium bromide. Quanti cation was performed using a two-step reaction process: reverse transcription (RT) and PCR. Each sample was run in triplicate for analysis. At the end of the PCR cycles, melting curve analysis was performed to validate the speci c generation of the expected PCR product. Quanti cation was performed using the 2 − ΔΔCT method, and data were normalized to the ACT2/7 transcript [41]. The sequences of the primers used are listed in Additional le 1: Table S1.

Statistical analysis
Signi cant differences were calculated using a one-way ANOVA analysis with a Turkey test and a signi cance level at p ≤ 0.05 and p ≤ 0.01 in SPSS 19.0 software [9]. All expression analyses were repeated three times. The reported value represents the arithmetic mean of three replicates. Data was expressed as a mean plus or minus a standard deviation (mean ± SD). Availability of data and material All data generated or analyzed for this study are included in this article and its supplementary information les.

Competing interests
The authors declare that they have no competing interests     The morphology of owers at different developmental stages in two Lonicera Japonica Thunb.