Volatile metabolome and transcriptome reveal fragrance release rhythm and molecular mechanisms of Rosa yangii

ZHOU Li-jun,HUANG Run-huan,LlU Ting-han,LlU Wei-chao,CHEN Yun-yi,Lü Pei-feng,LUO Le,PAN Hui-tang,YU Chao,ZHANG Qi-xiang

Beijing Key Laboratory of Ornamental Plants Germplasm Innovation & Molecular Breeding/National Engineering Research Center for Floriculture/Beijing Laboratory of Urban and Rural Ecological Environment/Key Laboratory of Genetics and Breeding in Forest Trees and Ornamental Plants of Ministry of Education/School of Landscape Architecture,Beijing Forestry University,Beijing 100083,P.R.China

Abstract Rose is a highly significant ornamental plant with substantial edible and medicinal value,cultivated worldwide primarily for perfume production. Recently,Rosa yangii,a new species found in northwestern Yunnan,China,has drawn attention due to its strong sweet scented flowers. In this study,the floral components of R.yangii were extracted at different flowering stages using solid phase micro extraction (SPME) and analyzed through gas chromatography-mass spectrometry (GC-MS). A total of 131 volatile organic compounds (VOCs) were detected from R.yangii,including 69 odor compounds. The production and release of floral VOCs were the highest during the initial-open stage,making it the most suitable time for harvesting as a significant number of floral components were synthesized and preserved.The analysis of the odor activity values (OAV) highlighted several key aromatic ingredients of R.yangii,such as eugenol,methyleugenol,benzeneacetaldehyde and phenylethylalcohol,heptanal,decanal,(E)-2-hexen-1-yl acetate,caryophyllene,and others. Metabolome and time-order gene co-expression networks (TO-GCN) revealed that VOCs and benzenoids/phenylpropanoids,along with associated genes,played a pivotal role in the overall floral regulatory network of R.yangii. MYB and bHLH were identified as the essential regulatory factors governing the regulation of eugenol synthase (EGS) and isoeugenol synthase (IGS),consequently influencing the sweet scent of R.yangii. The findings of this study provide a scientific foundation for enhancing fragrance through molecular breeding of ornamental plants.Furthermore,the study facilitated the development and utilization of this new plant’s essential oil material in various industries,including food storage,aromatherapy,cosmetic,and perfumery.

Keywords: rose,sweet fragrance,SPME-GC-MS,aroma VOCs,TO-GCN

1.lntroduction

Rose (Rosaceae,Rosoideae,RosaL.) is widely recognized throughout the world for its beautiful flowers and appealing scent (Jiaoet al.2022). For centuries,rose has been considered to be an important medicinal food homology plant with high edible and economic values (Hegdeet al.2022). It has a high content of polyphenols,vitamin C,E,B,and carotenoids,which have an antioxidant synergy effect (Fenget al.2023).In addition,rose is also an essential oil resource in the perfume,chemical,and cosmetic industries. It is reported that the floral volatile of rose has potential economic applications in improving memory,sedation,food storage,and flavor improvement (Zenget al.2019). Because of its physiologically-active substances and pleasant fragrance,rose also has commercial value for its utility in a wide range of consumer products in our society including flower cakes,jam,tea,essential oil,etc.(Deshusses 1947;Qinet al.2018).

The captivating aroma of roses,driven by volatile organic compounds (VOCs),is a key attribute that contributes to its high consumer demand (Xuet al.2022).A lot of plants with strong fragrances are promising sources to produce essential oils. Essential oils are volatile mixtures of hydrocarbons with several functional groups and useful properties including antioxidant activity,antibacterial properties,repellent activity,pleasure,relaxation,etc.(Shalitet al.2003). Roses exhibit a high diversity of flower fragrance (Raymondet al.2018) and therefore,rose fragrance consists of hundreds of volatile compounds with diverse biosynthetic origins,including terpenes,esters,and phenolic derivatives,but their amount varies among the varieties (Tholl and Gershenzon 2015). Rose Fragrance Institute of Japan has identified seven kinds of rose scents from scent analysis of thousands of roses (Yasuyuki and Miyako 2009). Among them,the Damask classic scent is widely known and has been extensively used as rose-charm (Zhouet al.2020b).The major aroma compounds of rose essential oil are citronellol,geraniol,linalool,phenethyl alcohol,eugenol,and rose ether,with stable nature and high antioxidative,antidepressant,and antibacterial activities,which can be used as a natural odorant (Koksalet al.2015).Rosa damascenais the most important rose species used to produce rose water,attar of rose,and essential oils in the perfumery industry. However,compared to the‘typical rose scent’ roses,the other rose odors are far less developed and utilized.

Finding new plant essential oil materials is crucial to bring diversity in fragrances and also important for its commercial and economic benefits. In the course of the evolution of rose scent,both the European and the Chinese progenitors have contributed distinctive scent characteristics. The major scent components of European roses include 2-phenylethanol and monoterpenes,whereas the volatiles emitted by Chinese roses contain high amounts of phenolic methyl ethers(Scallietet al.2008). China is the main distribution center of the genusRosa(Tanet al.2017). Since the early 19th century,tea-scented old Chinese garden roses were successively introduced into Europe and had a definite and lasting influence on the development of modern garden roses (Menget al.2011).Rosayangii,a newly discovered species originating from northwestern Yunnan,China,is characterized by its large shrub vines delightful sweet-scented flowers. This original variety presents new opportunities to develop new plant essential oils.However,its aroma components and the evolution pattern of VOCs during the flowering process have not been identified and reported yet.

Research on roses has made significant progress in terms of isolating and characterizing genes and enzymes involved in various scent biosynthetic pathways. These include germacrene D synthase (GDS),carotenoid cleavage dioxygenase (CCD),linalool synthase (LIS),alcohol acyl-transferase (AAT),1-deoxy-D-xylulose 5-phosphate synthase (DXS),1-deoxy-D-xylulose-5-phosphate reductoisomerase (DXR),nudix hydrolase(NUDX),phenylacetaldehyde synthase (PAAS),phenylacetaldehyde reductase (PAR),aromatic amino acid aminotransferase (AAAT),pyruvate decarboxylase(PPDC),eugenol synthase (EGS),phloroglucinol O-methyltransferase (POMT),orcinol O-methyltransferase(OOMT),caffeic acid O-methyltransferase (OMT),aromatic L-amino acid decarboxylase (AADC) and more(Shi and Zhang 2022). Additionally,several transcription factors (TFs) have been identified for the regulation of structural genes related to VOCs,notably,ODO1 (Verdonket al.2005),EOBI (Van Moerkerckeet al.2011;Spitzer-Rimonet al.2012),MYB4 (Colquhounet al.2011a),EOBII(Spitzer-Rimonet al.2010;Colquhounet al.2011b),PH4 in phenylpropanoids/benzenoids pathway,and WRKY1,MYC2 (Honget al.2012),basic/helix-loop-helix 4 (bHLH4)(Chuanget al.2018) in terpenoids pathway. PAP1 (Zviet al.2012) has shown potential as an upstream regulator of multiple metabolic pathways. However,only MYB1 has been associated with floral scent production in roses,although its extract functions remain unclear (Yanet al.2011). Despite these advancements,many aspects of floral scent synthesis in the roses remain largely unknown.Previous studies have primarily focused on individual biomolecules or specific genes in isolation,often neglecting the comprehensive understanding of physiological processes and gene regulation. Consequently,the regulatory networks that control the synthesis pathways of rose floral scents,including the synthesis pathways of rose floral scents and the regulation of their emissions,remain unclear.

The aim of this study was to investigate the natural release rhythm of VOCs throughout the flowering stages ofR.yangiiusing the headspacesolid phase microextraction (HS-SPME) coupled with gas chromatography-mass spectrometry (GC-MS).Additionally,the regulatory mechanisms governing floral scents inR.yangiiwere explored through TO-GCN analysis. This study also analyzed the main aromatic ingredients from the perspective of odor orientation to facilitate commercial perfume production processes at an industrial scale and improve their commercial value.

2.Materials and methods

2.1.Plant materials and aroma survey

Rosayangii,introduced and cultivated in the South Tropical Garden located in Kunming,Yunnan,China(24.86°N,102.98°E) was used in this experiment. These plants showed robust growth and abundant flower production from March to May every year. For this experiment,flower blooming was divided into five stages;young bud stage (S1),bud stage (S2),initial-open stage(S3),half-open stage (S4),and full-open stage (S5)(Fig.1). The tests were conducted with three biological replicates for each stage,using three plants. All sample collections were carried out between 10:00 to 11:00 in the morning in 2021.

Fig. 1 Plant materials and sampling stages. S1,young bud stage;S2,bud stage;S3,initial-open stage;S4,half-open stage;S5,full-open stage;L,leaf.

Fig. 2 Differential compounds between flowers and leaves of Rosa yangii. A,Venn diagram of differential kinds of odor volatile organic compounds (VOCs) emitted from flowers and leaves. B,pie chart of proportional VOCs with significant differences. C,Venn diagram of 33 compounds detected alone or significantly up-regulated in flowers. D,bar graph of up-and down-regulated of significantly different compounds compared between flowers and leaves. S1,young bud stage;S2,bud stage;S3,initial-open stage;S4,half-open stage;S5,full-open stage;L,leaf;F,flower.

Fig. 3 Major aromatic ingredients analyses. A,matrix bubble diagram of aroma contribution of odor volatile organic compounds(VOCs) with significant differences in flowers and leaves. B,odor evolution radar map of Rosa yangii during flowering process.S1,young bud stage;S2,bud stage;S3,initial-open stage;S4,half-open stage;S5,full-open stage;L,leaf.

Fig. 4 Gene functional enrichment analysis. A,GO enrichment. B,KEGG enrichment.

Fig. 5 Time-ordered gene co-expression network related to flower fragrance. A,predicted regulatory network and the connection among transcription factors (TFs) and structural genes involved in benzenoids/phenylpropanoids,terpenoids and fatty acid derivatives biosynthesis pathways. TF-O,other transcription factors. S1,young bud stage;S2,bud stage;S3,initial-open stage;S4,half-open stage;S5,full-open stage. B,expression heatmap of structural genes in benzenoids/phenylpropanoids pathway. PAR,phenylacetaldehyde reductase;AOC,primary-amine oxidase;PPDC,phenylpyruvate decarboxylase;AADC,phenylalanine decarboxylase;GOT,aspartate aminotransferase;hisC,histidinol-phosphate aminotransferase;PPA-AT,phenylpyruvate aminotransferase;PAL,phenylalanine ammonia-lyase;C4H,cinnamate-4-hydroxylase;C3H,P-coumarate-3-hydroxylase;4CL,4-coumaroyl-CoA ligase;COMT,caffeic acid 3-O-methyltransferase;CCoAOMT,caffeoyl-CoA O-methyltransferase;CCR,cinnamoyl-CoA reductase;CAD,cinnamyl alcohol dehydrogenase;CFAT,coniferyl alcohol acyltransferase;EGS,eugenol synthase;IGS,isoeugenol synthase;IEMT,(iso)eugenol O-methyltransferase;POMT,phloroglucinol O-methyltransferase;OOMT,orcinol O-methyltransferases.

Fig. 6 Sub-network for benzenoids/phenylpropanoids biosynthesis. A,sub-network for genes in benzenoids/phenylpropanoids biosynthesis. TFs,transcription factors. S1,young bud stage;S2,bud stage;S3,initial-open stage;S4,half-open stage;S5,fullopen stage. L1-L7,different levels. B,sub-network for eugenol synthase (EGS) and isoeugenol synthase (IGS).

For the sensory survey,a method previously employed in research (Zhouet al.2020a) was replicated. Seven rose varieties with standard scents,based on the seven kinds of rose fragrances from Rose Fragrance Institute of Japan (Yasuyuki and Miyako 2009) were selected.The sensory evaluation room,following the guidelines of international standard ISO 11136-2014 (2014) was carefully prepared. The room was clean,odor-free,noisefree and semi-ventilated room. In order to prevent any natural odor interference,approximately 3 g of flowers at each of the five stages along with the leaves from theR.yangiiwere collected and placed in clean water for 1 h within the sensory evaluation room,which had proper air circulation. The room temperature was maintained at 20°C and the relative humidity was within the range of 40-55%. To ensure accurate smell detection,the evaluators were trained according to guidelines provided in GBT15549-2022 (2022). The participants with sensitive olfactory sensation were recruited by sniffing and judging two cups of colorless liquid with alcohol and vinegar.Subsequently,seven participants (three females and four males with an average age of 23 years) were screened out as evaluators. They were tasked with smelling the samples and categorizing their fragrances by comparing them to the seven standard materials. They were also asked to score the intensity of floral fragrance for each sample ranging from 0 (no fragrance) to 100 (strong fragrance). The aroma survey was also conducted between 10:00 to 11:00 in the morning in 2021.

2.2.Extraction and determination of volatiles by SPME-GC-MS

Materials from the samples were harvested,weighted,and frozen immediately in liquid nitrogen,and kept in storage at -80°C. When needed,the samples were ground to a powder form in liquid nitrogen. Afterwards,a precisely measured 1.00 g of powder sample was weighed and immediately transferred to a 20-mL headspace vial (Agilent,Palo Alto,CA,USA),containing NaCl (analytical reagent) saturated solution,to prevent any enzyme reactions. To monitor the stability during testing and for the calculation of VOC content using a semi-quantitative method,50 μg mL-13-hexanone-2,2,4,4-d4 (10 μL,chromatographically pure) was added as an internal standard procedure (Chenet al.2020).The internal standards were prepared with hexane(chromatographically pure) and stored at -20°C. The vials were sealed using crimp-top caps with TFE-silicone headspace septa (Agilent). Each vial was placed in the oven at 100°C for 5 min. Afterwards,a 120-μm Divinylbenzene/Carboxen/Polydimethylsiloxane (DVB/CAR/PDMS) fiber (Agilent) was kept in the headspace of the sample for 15 min at 100°C.

The detection of volatiles was performed using the Agilent 8890-5977B platform at MetWare (www.metware.cn).The measurement parameters primarily followed Yuanet al.(2021). After sampling,desorption of VOCs from the fiber coating was performed in the injection port of the gas chromatography apparatus (Model 8890;Agilent) at 250°C for 5 min in splitless mode. VOCs were identified and quantified using an Agilent Model 8890 GC and a 5977B mass spectrometer (Agilent),equipped with a 30 m×0.25 mm×0.25 μm DB-5MS (5% phenyl-polymethylsiloxane)capillary column. Helium was used as the carrier gas at a linear velocity of 1.0 mL min-1. The temperature of the injector and detector was maintained at 250 and 280°C,respectively. The oven temperature was programmed to start at 40°C for 3.5 min,then increase to 100°C at a rate of 10°C min-1,increase to 180°C at a rate of 7°C min-1,and finally reach 280°C at a rate of 25°C min-1,which was then maintained for a total of 5 min. Mass spectra were recorded in electron impact (EI) ionization mode at 70 eV.The temperature of quadrupole mass detector,ion source,and transfer line were set at 150,230,and 280°C,respectively. Mass spectra were scanned in the rangem/z50-500 Da at 1 s intervals. The superposition diagram of the total ion current (TIC) diagram of quality control(QC) samples showed a high overlap of the TIC curves of metabolite detection,which meant the retention time and peak intensity were consistent and indicated that the reliable stability during the detection process of samples.In order to analyze the repeatability of samples under the same treatment method,the sample extracts were mixed as QC samples and inserted at intervals to monitor the stability of the instrument during detection.

2.3.Screening aroma volatiles and differential compounds

Qualitative analysis of the raw data obtainedviaGCMS was performed using Qualitative Analysis Workflow B.08.00 by comparing the mass spectra with the data system library (MWGC and NIST) and linear retention index (RI). Kovats retention indices for each compound were calculated using the linear formula ofn-alkanes(C7-C40) and were then compared with the theoretical retention index (Selliet al.2012). Quantitative analysis was performed using MassHunter Software (Agilent) for integrating and correcting chromatographic peaks (Xiaet al.2021). Peaks were filtered to retain only peak area data with a single set of null values not exceeding 50% or all sets of null values not exceeding 50%. The different samples were normalized according to internal standard procedures. Metabolites with significant differences in content were determined by a fold change ≥2 or ≤0.5,aP-value＜0.05,and variable importance in project (VIP) ≥1.

The selection of odor compounds was based on references from Pherobase Database (www.pherobase.com).Odor descriptions were sourced from TGSC Information System (www.thegoodscentscompany.com). The odor activity values (OAV),which represent the aroma contribution of each VOC,were calculated by dividing the relative content by the odor threshold. The odor threshold of 39 aroma components came from Gemert (2011). The OAVs of others were not calculated either because they were not found or odorless.

2.4.RNA sequencing and differentially expressed genes (DEGs) analysis

RNA sequencing (RNA-seq) data were generated from three biological replicates that were used for each treatment group (S1,S2,S3,S4,S5 and L). The quality of extracted RNA was assessed using 1% agarose gels.The NanoPhotometer?spectrophotometer (IMPLEN,CA,USA) was used to check the RNA purity. RNA concentration was measured using Qubit?RNA Assay Kit in Qubit?2.0 Flurometer (Life Technologies,CA,USA). RNA integrity was assessed using the RNA Nano 6000 Assay Kit of the Bioanalyzer 2100 System (Agilent Technologies,CA,USA).

The cDNA libraries were prepared using the Illumina sequencing platform by Metware Biotechnology Co.,Ltd.(Wuhan,China) and sequenced on the NEBNext?UltraTM RNA Library Prep Kit for Illumina?(NEB,USA).The TruSeq PE Cluster Kit v3-cBot-HS (Illumia) was used for clustering the index-coded samples on a cBot Cluster Generation System. Subsequently,the library preparations were sequenced on an Illumina Hiseq platform,generating 125 bp/150 bp paired-end reads.

Fastp v0.19.3 was employed to remove reads containing adapters or with an N content exceeding 10%of the base number. Additionally,reads with more than 50% of the low-quality bases (Q≤20) were filtered out.The resulting clean reads were mapped to theR.chinensis‘Old Blush’ reference genome (https://www.ncbi.nlm.nih.gov/data-hub/genome/GCF_002994745.2/) (Raymondet al.2018) using HISAT v2.1.0. This allowed the construction of an index and comparison of clean reads to the reference genome (Kimet al.2015).

A total of 18 RNA-seq libraries were obtained,resulting in 130.82-Gb clean data after filtering out low quality reads were acquired. More than 87% of the reads in each library could be uniquely mapped to the availableR.chinensis‘Old blush’ reference genome. However,due to poor repeatability in the biological replicates,the S2-1 sample was excluded from further consideration based on consideration by sample correlation analysis (Appendix A).

StringTie v1.3.4d was employed to predict new genes,while FeatureCounts v1.6.2 was used to calculate the gene alignment and determine the Fragments Per Kilobase of exon model per Million mapped fragments(FPKM) for each gene based on its length. DESeq2(Loveet al.2014) was applied to identify DEGs. Pairwise comparisons were performed between the five flower sample groups and leaf samples. DEGs were filtered based on fold change values ≥2 and the Benjamini-Yekutieli false discovery rate with FDR≤0.01 using Tbtools. Among the 42 141 expressed genes,22 288 genes showed DEGs during flowering progress compared with the leaves (Appendix B).

2.5.Time-ordered gene co-expression network analysis (TO-GCN)

The construction of TO-GCN followed the methods described by Chang (2019) and Xu (2021). In this study,a total of 33 compounds were either detected alone or significantly up regulated in flowers along with 22 288 DEGs,including 437 TFs and 21 851 structural genes. To establish the TO-GCN,Pearson’s correlation coefficient(PCC) values were calculated for each pair of VOCs and genes. Correlations under the mode of ‘C1+C2+’ were selected to generate hierarchical levels in the TO-GCN.The PCC cutoff value between TFs and structural genes was set at 0.95,while for VOC-gene correlations,it was set at 0.80. Pairwise nodes with low PCC values were filtered out,resulting in the construction of a preliminary network,which comprised 332 TFs,12 903 structural genes and 33 VOCs. The preliminary network consisted of major GCN (including 294 TFs,12 798 structural genes and 33 VOCs) and other 34 scattered GCNs including the remaining 38 TFs,and 104 structural genes,which were filtered out in the subsequent steps. An ethyleneresponsive transcription factor WIN1 (LOC112190153)was chosen as the initial node for the network due to its high expression at S1 and subsequent down-regulation in all other stages. The presence of several proteins in the ERF family,which are known to play diverse functions in developmental processes in various plant species,supported our hypothesis that WIN1 is involved in the flowering process. The breadth-first search (BFS)algorithm was applied to assign the time-ordered levels for all TFs,structural genes and VOCs in the major GCN.The process generated nine time-ordered levels of nodes,indicating the inferred expression time order during flowering development for all genes in the major GCN.

2.6.Functional enrichment analysis and transcription factor identification

The Unigenes obtained in this study were used as query sequences for annotation against various databases,including the NCBI non-redundant (NR) (http://www.ncbi.nlm.nih.gov/) protein sequences and Swiss-Prot (http://www.ebi.ac.uk/uniprot/),Gene Ontology (GO) (http://www.geneontology.org/),Kyoto Encyclopedia of Genes and Genome (KEGG) pathways (http://www.genome.jp/kegg/)and the Pfam (Protein family) (http://pfam.sanger.ac.uk/)databases.

To determine whether specific functional GO term and KEGG pathway categories were significantly overrepresented in specific gene sets compared with the background set of all annotated genes in theR.chinensisreference genome,hypergeometric tests were performed using Tbtools (Chenet al.2020). Each gene set corresponding to the initial,transitional and terminal stages in the TO-GCN was analyzed. Significantly enriched GO terms and KEGG pathways were identified based on an adjustedP-value threshold of ≤0.05. For GO terms,those with aP-value ≤0.001,enrichment score ≥2.00 and hits genes counts in selected set ≥20 were considered significant. KEGG pathways related to metabolism were shown. The Unigene sequences were categorized into biological process (BP),cellular component (CC),and molecular function (MF) aspects of GO terms. Transcription factor sequences were translated into protein sequences using Softberry(http://www.softberry.com/),and phylogenetic trees were constructed with transcription factor family protein sequences ofArabidopsisthaliana(http://planttfdb.gaolab.org/) using the Maximum Likelihood (ML) method.

3.Results

3.1.Specific odor compounds of flowers in R.yangii

The sensory evaluation results for the aroma type ofR.yangiidemonstrated that it emitted a pronounced sweet fragrance,which was misreported as a tea-scent.In fact,this distinctive aroma is a characteristic Chinese rose. The aroma of roses is due to a combination of various volatile components. A total of 131 VOCs more than 0.01 μg g-1of sample were detected by SPME listed in Appendix C,including 33 terpenoids,32 hydrocarbons,16 alcohols,11 aldehydes,11 heterocyclic compounds,10 esters,eight aromatics,five ketones,four phenols,and one ether. Previous research has shown that only a few compounds are actually involved in producing the distinctive aroma of several foods (Polaskovaet al.2008). Long-chain alkanes are common unscented compounds that are found in many kinds of plant extracts and essential oils but the long-chain alkane matrix had no significant contribution to the olfactory threshold of aroma matrix (Fuet al.2022). Therefore,it is important to identify only those compounds that are dominantly contributing to an odor. In this paper,several odorless volatile compounds,such as alkanes-like dodecane,tetradecane,olefin-like pentadecen,heptadecene,etc.,which made an insignificant contribution to the floral fragrance were excluded,following proper verification from literature and relevant websites. Consequently,a total of 69 odor compounds (Appendix D) were screened and selected from 131 VOCs. These odor volatiles were mainly grouped into three biochemical synthesis pathways,including 30 terpenoids,23 fatty acid derivatives,14 benzenoids/phenylpropanoids,and two others.

Compared to the odor of VOCs from the leaves,the flowers release nine unique aromatic compounds (Fig.2-A),including methyleugenol,isoeugenol,4′-hydroxyacetophenone,(+)-epibicyclosesquiphellandrene,epizonarene,α-cadinene,2-methyl-1-butanol,hexadecanol,and tetradecanal.The sixty other odorant compounds were detected in both flowers and leaves,among which 31 compounds had no significant difference,24 compounds were significantly up-regulated in flowers,and five compounds were significantly up-regulated in leaves (Fig.2-B and D). Above all,33 compounds that were detected alone or significantly up regulated in flowers were more likely to have a greater impact on the aroma ofR.yangii. Four compounds were significantly different from leaves at every stage of flowering,which were eugenol,methyleugenol,(E)-isoeugenol,and tetradecanal(Fig.2-C). Interestingly,the first three compounds were produced from the same synthetic pathway.

3.2.Major aromatic ingredients and the dynamic changes of fragrance in R.yangii

There were very minute differences in the type of VOCs at different flowering stages;however,the relative amount for VOCs with different odor threshold values varied significantly between different stages. The matrix bubble diagram illustrates the different contributions of VOCs to floral fragrance with significant differences in leaves and flowers according to log10(OAV) as Fig.3-A.Eugenol showed a dominant state with not only high content but also low odor threshold. Additionally,methyleugenol,the methylated product of eugenol,as one of the five unique components detected only in flowers,also produced a distinct effect on the fragrance from initial to full open stage. The other four unique components were not easily detected by humans with higher odor thresholds,and contributed little to floral aroma. Moreover,benzeneacetaldehyde and phenylethyl alcohol made significant contribution to the fragrances ofR.yangii. Heptanal,decanal,(E)-2-hexen-1-yl acetate,caryophyllene,etc.,also played a role in the overall fragrance.

Fragrance is a highly complex component of floral phenotype,with dynamic patterns of emission and chemical composition (Raguso 2008). In Fig.3-B,the eight aromatic ingredients with OAV ＞1 of the fragrance were presented. Overall,throughout the flowering process,the emissions of the majority of odor VOCs increased from the stage of young bud to initial-open flowers stages and decreased from half-open to fullopen flowers stages. This trend aligns with the observed changes in aroma intensity as detected by sensory evaluation (Appendix E). The results show that the most suitable harvest time was near the initial opening to halfopen stage,when a large number of floral components were synthesized and preserved in the flowers.

3.3.Overall gene expression patterns in R.yangii throughout the flowering progress

By matching the expression time order of DEGs across the five flowering stages and aligning them with the assigned levels of TO-GCN,we identified a distinct developmental-stage transition during the scent-releasing process. The transition was observed in the initial stage(corresponding to S1 and L1-L2),the transitional stage(corresponding to S2 and L3-L4),and the terminal stage (corresponding to S3-S5 and L5-L9) of fragrance emission (Appendix F). This pattern resembles the corolla pigmentation observed in rhododendron (Yanget al.2020).

GO functional enrichment analysis revealed that distinct gene functions were activated or de-activated during the flowering process (Fig.4-A). In the initial stage,a significant number of DEGs were enriched in GO terms such as ‘RNA modification’,‘RNA binding’,and ‘structural molecule activity’,among others. During the transitional stage,‘pollination’ and ‘multi-multicellular organism process’ were overrepresented. At the terminal stage,all DEGs were significantly enriched in Biological Process categories,such as ‘regulation of abscisic acid-activated signaling pathway’ and ‘negative regulation of response to stimuli. Considering the intricate biological activities in plants,some of these processes are associated with the synthesis and release of VOCs. Certain compounds originating from specific biosynthetic pathways play a crucial role in mediating interactions between flowers and their pollinators (Farre-Armengolet al.2020).

When the kids came in, he took them for walks along the pier21 near their office. Often she went along and watched Eric, who was becoming a master of sign language, talk and laugh with her boys as no one else had before.

We also conducted KEGG enrichment analysis to investigate the dynamic characteristic of the transcriptome inR.yangii(Fig.4-B). The majority of DEGs were enriched in the Metabolism Class. During the initial stage,numerous metabolism pathways were activated in young buds to prepare for flowers blooming,including‘Glycolysis/Gluconeogenesis pathway’,‘Starch and sucrose metabolism’ and others. At the terminal stage,DEGs were enriched in pathways such as ‘Carbohydrate metabolism’ and others.

Regarding the synthesis of aroma VOCs,pathways related to fatty acid biosynthesis and fatty acid elongation were enriched during the initial stage,while fatty acid degradation was enriched at the terminal stage.Phenylalanine metabolism pathways were enriched at the initial stage,whereas phenylpropanoid biosynthesis pathways were enriched during the transitional stage.Pathways associated with terpenoids such as terpenoid backbone biosynthesis,were enriched at the terminal stage. By examining the significant enrichment of KEGG pathways related to floral volatile synthesis at different stages,we observed that compounds from various biosynthesis pathways were synthesized in a specific order as the flowers bloom.

3.4.Time-ordered fragrance-related gene coexpression in R.yangii

To reveal the regulatory mechanism of fragrance during the flowering process inR.yangii,we identified 222 genes encoding enzymes functioning in the benzenoids/phenylpropanoids,terpenoids and fatty acid derivatives biosynthesis pathways. A sub-TO-GCN consisting of seven levels was extracted to reveal a co-expression network that involves 294 differentially expressed TFs and 222 structural genes related to scent-related biosynthesis pathways (Fig.5-A). The network comprised 516 nodes(Appendix G) and 23 711 edges (Appendix H). Within the network,six transcription factor families known to be associated with aroma compound synthesis pathways were represented by colored dots,including MYBs,NACs,bZIPs,ERFs,WRKYs,and bHLHs. Other transcription factors were indicated by grey dots. The overall network showed that the majority of genes (284,including 167 TFs and 117 structural genes) exhibited peak expressions at the initial stage,indicating subsequent down-regulation during flowering progress. Of the remaining genes,125(78 TFs,47 structural genes) were at the transitional stage,while 107 genes (49 TFs,58 structural genes) were observed at the terminal stage.

The analysis of the gene regulation dynamics in three pathways revealed that structural genes involved in benzenoids/phenylpropanoids (represented by blue nodes) had the most significant impact on the overall floral regulatory network. This observation was consistent with the analysis of aromatic components ofR.yangii. The majority of genes encoding key biosynthetic enzymes in aroma determination exhibited high expression levels at S1 and S2,including those involved in the phenylpropanoid biosynthesis pathway encoding eugenol synthase,isoeugenol synthase,and others. In contrast,the related VOCs displayed a delay in reaching high expression S3,indicting a time lag from transcriptional regulation to the synthesis and release of VOCs (Fig.5-B). This finding suggests a temporal gap between transcriptional variation and the accumulation of metabolites during flowering.

3.5.Regulatory subnetworks for major aroma VOCs of R.yangii

Specifically,during the initial stage of flowering,a total of 64 enzymatic genes involved in benzenoids/phenylpropanoids biosynthesis exhibited high expression(Fig.6-A). Among the 98 identified TFs potentially regulating these enzymatic genes,the bHLH and MYB factors were most abundant. At the transitional stage,WRKY and ERF family members played a significant role in regulating the genes. However,at the terminal stage,the number of TFs with high expression decreased.

In the case ofR.yangii,eugenol and isoeugenol were identified as major aromatic compounds contributing to the sweet fragrance. Three genes (LOC112168914,LOC1121717310 and LOC112201866) were discovered to play an important role in the synthesis of (iso)eugenol.Previous studies have reported numerous transcription factors involved in the synthesis of floral volatiles from the benzenoids/phenylpropanoids pathways,including MYB transcription factors,such as PhEOBII to activate PhODO1 and IGS (isoeugenol synthase) in petunia and FaMYB10 could up-regulate the expression of FaEOBII and FaEGS2 and promote the production of eugenol(Spitzer-Rimonet al.2010;Colquhounet al.2011a;Van Moerkerckeet al.2011;Medina-Pucheet al.2015).However,the upstream regulators of EGS and IGS in roses have yet to be identified.

ForR.yangii,a total of 78 TFs were identified as regulators of EGS and IGS,with majority belonging to the bHLHs (32) and MYB (22) families. This was determined by analyzing the co-expression network inferred from TOGCN (Fig.6-B). Among these TFs,three bHLHs were found to be associated with both two RyEGSs and RyIGS.Additionally,a significant number of transcription factors(13 bHLHs,nine MYBs,three ERFs,two WRKYs and a NAC) were identified as potential regulators of EGS,and some TFs were found to influence both one EGS(LOC112171731) and IGS. However,no TFs were found to be associated with the other EGS (LOC112168914).Moreover,these three genes also exhibited self-regulation by specific transcription factors.

By examining the network topology,seven TFs were identified as the hub genes,including three MYBs,three bHLHs and a bZIP TFs,which were represented by rhombus shapes. Furthermore,based on the previous reports,we identified the MYB gene family and identified nine R2R3-type MYBs,including LOC112182886,LOC112195924,LOC112200243,LOC112173673,LOC112175858,LOC112200214,LOC112175573,LOC112172309 and novel.2237. These MYBs were represented by red lines connecting them to structural genes. Several R2R3-type MYBs have been reported as positive regulators of structural genes involved in floral volatile synthesis in benzenoids/phenylpropanoids pathways,such as ODO1,EOBII,MYB4,EOBI and PH4.

4.Discussion

The advancement of sequencing technology enabled the integration of transcriptomics and metabolomics allowing for rapid analysis of trait formation mechanisms.This approach has been extensively applied in various plant research,such asCoriandrumsativum(Wuet al.2021) andPassifloraedulis(Xuet al.2023). However,it is important to note that there is a time lag between gene expression and the formation of phenotypes.Consequently,conventional methods relying on the consistency of genes and metabolites expression trends may lead to inaccuracies in screening related genes. Time-shift effects of plant metabolites and gene expression have been observed in various plants. For example,Zhanget al.(2016) reported systematic transcriptome and metabolites shifts modulated inIsatis indigoticahairy roots,while Changet al.(2019) revealed time-shifted expression profiles and developed TOGCN to uncover the master regulators of entire signaling pathways by integrating data from RNA sequencing and physiological traits.

The approach of avoiding time shift effects has proven to be more appropriate in studying potential regulatory networks of timing and coloration inRhododendronsimsii(Yanget al.2020),R.molle(Nieet al.2022),Syringa oblata(Maet al.2022),among others. However,since the genome of sweet-scented rose has not been published,it is not possible to predict the promoter sequence binding sites of related genes in this study. Therefore,the regulatory mechanisms underlying the formation of VOCs require further experimental verification.

Additionally,it is worth noting shikimic acid pathway served as the upstream pathway for the synthesis of benzenoids/phenylpropanoids volatile compounds and anthocyanins. In the case ofR.yangii,as flowers release their scent,the colors also undergo a transition from white with light yellow and red streaks to a deeper red in the afternoon. Some of the potential TFs predicted in this co-expression network of eugenol synthesis have been reported to regulate flavonols and anthocyanins.These TFs include genes such as LOC112200214,LOC112200243,LOC112171864,LOC112167662,LOC112173463,LOC112174143,LOC112190178,LOC112186010 and others. The co-regulation of flower color and fragrance ofR.yangiipresents an intriguing avenue for further research.

5.Conclusion

This paper investigated the floral scents ofR.yangiifound in wild for the first time. The study identified rich functional floral compounds by volatile metabolomic detection,which provides strong evidence for economic potential of essential oil development. A total of 131 volatile compounds were identified and 61 odor compounds were screened out from floral components. Flowers released a lot more aromatic compounds in benzenoids/phenylpropanoids biosynthetic pathway as compared to the leaves. The main aromatic ingredients ofR.yangiiincluded eugenol,methyleugenol,benzeneacetaldehyde and phenylethylalcohol,heptanal,decanal,(E)-2-hexen-1-yl acetate,caryophyllene,etc.,which have their useful applications in food storage,aromatherapy,cosmetic and perfumery industries. Furthermore,sensory evaluation and metabolite detection results showed that initial-open stage was more suitable for harvesting the flowers at an industrial scale due to the fact that a large number of floral components were synthesized and preserved in the flowers during this stage. In addition,the analyses of TO-GCN revealed the hierarchical networks of major components of sweet fragrance inR.yangii. The bHLH,MYB,WRKY,ERF,NAC and bZIP transcription factors were the major regulatory elements of the key biosynthetic genes controlling VOCs metabolism. MYBs and bHLHs played an important role in regulating benzenoids/phenylpropanoids biosynthesis during flowering development. The enzymatic genes and potential regulators of aroma VOCs biosynthetic pathways represent worthy priorities to further functional research about flower fragrance in roses.

Acknowledgements

This work was supported by the Beijing High-Precision Discipline Project,Discipline of Ecological Environment of Urban and Rural Human Settlements and the National Key R&D Program of China (2019YFD1000400). Special thanks to Mr.Yang Yuyong (Kunming Yang Chinese Rose Gardening Co.,Ltd.,China) for the kind assistance in collecting and sharingRosagermplasm resources,Mr.Zhao Shiwei and Mr.Nie Shuai (Beijing Forestry University,China) for the guide in the construction of TOGCN. In addition,thanks for MetWare (www.metware.cn)providing the testing platform.

Declaration of competing interest

The authors declare that they have no conflict of interest.

Appendicesassociated with this paper are available on https://doi.org/10.1016/j.jia.2023.06.015