Transcriptome analysis reveals regulatory networks and hub genes in flavonoid metabolism of Rosa roxburghii

Rosa roxburghii Tratt, the most popular fruit that blooms in the southwest of China, is rich in flavonoids. However, the regulatory network and critical genes involved in the metabolism of flavonoid compounds in R. roxburghii are still unknown. In this study, we revealed that flavonoid, anthocyanin and catechin accumulated at different levels in various tissues of R. roxburghii. We further obtained and analyzed differentially expressed genes (DEGs) involved in flavonoid metabolism from five samples of R. roxburghii by transcriptome sequencing. A total of 1 130 DEGs were identified, including 166 flavonoid pathway biosynthesis genes, 622 transcription factors, 301 transporters, and 221 cytochrome P450 proteins. A weighted gene co-expression network analysis (WGCNA) of the DEGs uncovered different co-expression networks. In terms of biosynthesis enzymes, cytochrome P450 CYP749A22 and CYP72A219 were highlighted in regulation flavonoids content. Anthocyanin 3-O-glucosyltransferase and F3’H were the top two critical enzymes for anthocyanin content. In contrast, caffeic acid 3-O-methyltransferase, 4-coumarate-CoA ligase, and shikimate O-hydroxycinnamoyl transferase were essential for catechin accumulation. Additionally, the eigengene network of the “black” module had high correlations with total flavonoid (r= 0.9, p=5e-06). There were 26 eigengenes in the “black” module, including six flavonoid biosynthesis, 14 TFs and six transporters. Among them, besides cytochrome P450 proteins (DN136557_c0_g1, DN135573_c0_g1 and DN145971_c4_g1), isoflavone-hydroxylase (DN143321_c3_g1) was crucial for total flavonoids content based on the high degree of connectivity. The transcription factors RrWRKY45 (DN142829_c1_g5), RrTCP20 (DN146443_c1_g1) and RrERF118 (DN141507_c3_g2) were significantly correlated with flavonoids in R. roxburghii. The present transcriptomic and biochemical data on metabolites should encourage further investigation on functional genomics and breeding of R. roxburghii with strong pharmaceutical potential.


132
Five different samples were applied to high-throughput Illumina sequencing and been published [35].

133
Tissue collection and preparation were performed as previously described [35]. The reads were De novo 134 assembly using the Trinity with default settings based on the de Bruijn graph algorithm [31]. All clean 135 reads generated by Illumina sequencing have been deposited, being publicly available and can be readily  [30]. The WGCNA network and module were conducted using an unsigned type of topological overlap 152 matrix (TOM). The calculation parameters "soft thresholding power" = 2 and "merge Cut Height" = 0.5 153 were selected to analysis of the DEGs. Finally, the module eigengene value was calculated to evaluate 154 the relationships among modules, total flavonoid, anthocyanin and catechin content in the five samples.

155
of modules with each tissue type.

156
Identification of hub genes.

157
The kME was determined as the Pearson correlation coefficient between each gene. The module eigengene was used to evaluate the association of the module with total flavonoids contents. The most 159 significant module ('black') of genes with WGCNA edge weight >0.80 was represented using Cytoscape  Table S7. To ensure 172 the reproducibility of results, we carried out quantitative PCR analysis in triplicate for each sample. β-173 actin was used as an internal control. The value of a 2 -ΔΔCt method was used to assess the relative gene 174 expression.

176
The data were expressed as mean ± standard deviation (SD) and analyzed using analysis of variance 177 (ANOVA) and SPSS (Statistical Package for Social Sciences) Statistical 20.0. Statistical significance 178 was set at p < 0.05.

180
Quantification of Total Flavonoids, Anthocyanin and catechin in different tissues.

181
The previous studies proved that the leaves and fruits of R. roxburghii are rich in total flavonoids and 182 catechin [4,5,44]. In this study, the tissues from stems, leaves, flowers, young fruits, and mature fruits 183 were further used to investigate the distribution of total flavonoids in R. roxburghii. The flavonoid was 184 present in all tissues (Fig. 1). Their content was highest in mature fruits (237.03 mg/100 g FW), followed 185 by young fruits, flowers, and leaves, with the contents between 70.27-175.87 mg/100 g FW. Less than 186 45 mg/100 g FW in stems of mature seedlings. However, anthocyanins were highly accumulated in 187 flowers of R. roxburghii. The content of anthocyanin in flowers was 5.63 mg/100 g FW, followed by the other four tissues, ranging from 2.27 -0.90 mg/100 g FW. Moreover, we determined the content of 189 catechin in five samples. The values varied from 8.47 to 1.10 mg/100 g FW with a descending order of 190 leaves, mature fruits, young fruits, flowers, and stems, confirming that the leaves were highly rich in 191 catechin. These results indicated that total flavonoids and specific components of flavonoids could 192 accumulate in various tissues of R. roxburghii at relatively high concentrations. Therefore, to decipher 193 the hub genes which regulated these flavonoids, especially the accumulation of total flavonoid, we 194 performed transcriptome sequencing from various tissues in this study.

195
Analysis of differentially expressed genes.

196
To preliminarily explore gene expression and then analyze the genes that regulated flavonoids in different 197 tissues, we determined a total of 25 449 DEGs between any two tissues based on a corrected adjusted P-

206
Based on GO, KEGG and the Nr library comparison using BLASTx alignment, we evaluated enzyme-207 encoding genes of the flavonoid pathways (Table S1). The analysis of transcriptome data uncovered that 208 166 key candidates exerted direct influence over twenty enzymes that were known to be involved in    (Table S3). The expression levels of transporters,

233
including ABC transporter, SNARE, GST, and MATE were shown in Figure 3A. However, most of the 234 candidate ABC transporters and GST displayed no expression in five samples.

235
DEGs annotated as transcription factors involved in flavonoid metabolism.

236
As many transcription factors (TFs) regulate the expression of flavonoid synthesis and transport, we also 237 identified putative TFs based on the transcriptome. In our results, a total of 622 unigenes were determined 238 as putative TFs (Table S4). The number of unigenes encoding MYB (128 DEGs) was the highest.  Most TFs were highly expressed in five samples, implying that they may take part in more metabolic 243 processes of R. roxburghii compared with transporters ( Fig. 3B),

244
The P450 Family Gene expression.

245
P450 enzymes involved in the generation of flavonoids [49]. By searching the DEGs, 221 DEGs were 246 annotated as putative cytochrome P450 members and could be grouped into CYP subfamilies (Table S5).

254
To investigate the regulatory network and determine the key genes that influenced flavonoids in R. anthocyanin and catechin accumulation, respectively. The genes in these modules were listed (Table S6).

267
The results showed that different gene expressions played roles in specific flavonoids content.

268
We selected the modules with correlation coefficient values ≥ 0.83 and analyzed eigengenes related

277
There were 26 eigengenes in the "black" module, including six flavonoid biosynthesis, 14 TFs and 6 transporters. To explore the critical genes that influences flavonoids content, we used Cytoscape software 279 to visualize the network of the genes in "black" modules ( Fig. 6A). Cytoscape representation of the 22 280 genes with WGCNA edge weight >0.80 indicated that these genes were highly positively connected in 281 the "black" module. In the interaction network diagram, the outer layer consists of 12 genes related to 282 TFs, including RrWRKY45, RrTCP20 and RrERF118. In the middle of the network diagram, four 283 flavonoid transporter genes were identified. The transcripts of 6 genes related to biosynthesis were 284 identified in the network. All genes in the "black" module showed high expression in mature fruits based 285 on transcriptome (Fig. 6B), further illustrating that the black module was very relevant to flavonoids 286 content.

287
Confirmation of the hub genes using qRT-PCR.

288
Based on ranking the connectivity of each node, we identified hub genes in the "black" module (edge ≥ 289 14). The expression levels of hub genes were determined by qRT-PCR using β actin as an internal control.

290
We could obtain that the transcription levels of DEGs found in module 'black' were nearly consistent 291 with the gene expression profiles obtained from RNA-seq (Fig. 7), indicating the reliability and accuracy

300
The transcriptome has become an effective tool to study biosynthesis mechanisms in plants.

301
Transcriptome analysis for a flavonoid investigation in apple yellow fruit somatic mutation has been flavonoids contents in mature fruits of R. roxburghii were higher than other tissues (Fig. 1). Therefore, 320 they were selected as three trait data for module-trait relationship analysis.

321
We identified the 166 DEGs that encoded the known enzymes involved in the biosynthesis of the and alterations in fruit color, the expression of CHS was increased [56,57]. We analyzed the expression 328 pattern of these DEGs related to RrPAL and RrCHS (Fig. 2). Strikingly, we found that most of them were                                        Table S7. The list of all genes in "black", "purple", "lightgreen", "blue", "green" and "yellow" modules.