Introduction

Plants produce a variety of volatile compounds and emit them as floral scents from the surface of leaves or flowers. The volatile compounds emitted from flowers play an important role for attracting pollinators or natural enemy of insects, deterring insects or herbivores, etc. (Girling et al. 2013;Klahre et al. 2011;Muhlemann et al. 2014;Zhang et al. 2016). Plant volatile compounds have very high volatility and vapor pressure at low temperature and are categorized into four large groups such as; terpenoids, aliphatics, benzenoids/phenylpropanoids, C₅ branched-chain compounds, nitrogen or sulfur containing volatiles, etc. (Bouwmeester et al. 2019;Dudareva et al. 2013).

Metabolomics is the research field to analyze all possible metabolites to clarify biological phenomena. To measure the levels of metabolites contained in a plant comprehensively, but since there is no means to completely cover all metabolites with different physical properties with a single instrument, various type of analytical instruments such as nuclear magnetic resonance (NMR), liquid chromatograph mass spectrometer (LC-MS), capillary electrophoresis mass spectrometer (CE-MS), gas chromatograph mass spectrometer (GC-MS) are used (Fukusaki and Kobayashi 2005;Kumar et al. 2017). GC-MS has a limit to analyze metabolites with low molecular weight ranging 50 - 500 m/z (mass to charge ratio), but it is powerful tool enough to analyze volatile organic compounds in plants and, therefore, this approach has been widely used to understand the composition of volatile compounds and temporal and spatial change in plants (Barman and Mitra 2019;Dötterl and Jürgens 2005). Recently, the head-space solid phase micro extraction (HS-SPME) technique coupled with GC-MS is increasingly used to analyze volatile compounds in plants because HS-SPME is easy to handle and also known to provide very high accuracy and reproducibility for routine analysis using autosampler (Arthur and Pawliszyn 1990;Dong et al. 2007).

Chrysanthemum boreale and C. indicum are wild Chrysanthemums native in Korea and also widely distributed to East Asian countries. In addition, C. indicum var. albescens has a white flowerhead and is known as a variety of C. indicum which has yellow flowerhead like C. boreale. C. boreale contains camphor, cis-chrysanthenol, α-thujene, 1,8-cineole, α-pinene, β-caryophyllene, germacrene D, camphene, umbellulone, and β-pinene (Hong 2002). On the other hand, the majority of volatile compounds of C. indicum is accounted by camphor and known to have very high biological activity (Youssef et al. 2020). It also contains α-pinene, 1,8-cineole, α-thujone, terpinen-4-ol, bornyl acetate, borneol, cis-chrysanthenol, β -caryophyllene, germacrene D, and α-cadinol (Jung 2009). The composition and activity of volatile compounds in wild Chrysanthemums has been widely studied. However, few studies have been reported on the change or improvement of volatile compounds in Chrysanthemums, but it is inevitable to deepen understanding of the profile of volatile compounds not only to improve both in quality and in quantity but also to find out the optimum harvesting period for a particular compound.

Therefore, in the current study, metabolomics approach using HS-SPME-GC-MS and multivariate analysis was conducted to understand temporal variation of volatile compounds of three wild Chrysanthemums according to four different flowering stages and also to identify the volatiles affected the difference of volatile compound profiles among three wild Chrysanthemums according to different flowering stages.

Materials and Methods

Plant material and sample preparation

Three wild Chrysanthemums, C. boreale, C. indicum, and C. indicum var. albescens were grown at the experimental field of Chung-Ang University, Anseong, Republic of Korea (37°00'02.4"N 127°13'41.1"E). Flowerheads of three Chrysanthemums were collected at the end of October 2011 according to the four different flowering stages. Flowering stages were determined depending on developmental and morphological traits of flowerheads; Before Flowering (BF, disc flowers are covered by yellowish green ray florets), Begin to open (BO, ray florets are slightly open and disc flowers are appears), Half open (HO, ray florets are open and their vertical angle is 45°), and Fully open (FO, ray florets are fully open and their vertical angle is 90°). Flowerheads of the same stage were pooled and then immediately immersed in liquid nitrogen. Samples were vacuum freeze-dried at -50°C for 72 h, Automill (Tokken, Inc., Japan) was used to grind samples at 1,000 rpm for 1 min, and then stored in a screw top vial (C4020-18, Thermo Scientific, Waltham, USA) with desiccants at -80°C until use. Samples for analysis were prepared by Tikunov et al. (2005) with slight modification. 50 mg of sample powder were added to a 20 mL headspace screw vial. 500 μL of 100 mM EDTA-NaOH and 1.5 g of CaCl₂·2H₂O were added. Finally, the vials were sonicated at room temperature for 5 min.

Sample vials were placed to the Multi-Purpose Sampler (MPS 2L-XT, GERSTEL, Germany) equipped to gas chromatographymass spectrometry (Agilent GC-MSD system, GC; 7890A, MSD; 5975C, Agilent, Santa Clara, USA) and agitated at 65°C for 20 min. The SPME fiver was thermally cleaned prior to absorbing volatile compounds at 250°C for 2 min in GC injection port (pre-injection). Finally, volatile compounds were absorbed using HS-SPME method using PDMS/DVB (polydimethylsiloxane/ divinylbenzene) fiber for 20 min. Volatile compounds were thermally desorbed in GC splitless injection port and then the fiber was thermally cleaned at 250°C for 2 min again before absorbing next sample (post-injection).

Measurement of volatile compounds of using HS-SPME -GC-MSD

Analysis of volatile compounds was performed using Agilent GC 7890A gas chromatograph coupled with mass spectrometer (5975C MSD, Agilent). HP-5MS (30 m × 0.25 mm × 0.25 μm film thickness, Agilent), a nonpolar capillary column, was used to separate volatile compounds. Injection temperature was set up to 250°C. Initial temperature was set up to 40°C for 1 min and then the temperature was raised to 250°C increasing by 5°C every minute. Interface line temperature of mass spectrometer was set up to 280°C. Mass spectra of volatile compounds were detected using scan mode with EI method (Electron Impact Ionization, 70.0 eV).

Data Analysis

m/z (mass to charge ratio) and retention index of volatile compounds were used to identify chemical name by comparing NIST05 Mass spectral library database (NIST, Gaithersburg, USA) and further checked using NIST Chemistry Webbook (https://webbook.nist.gov/chemistry/). Peak area of volatile compounds was measured by area normalization method using MSD Chemstation (Version E.02.02.1431, Agilent Technologies, Inc.). Retention Index (RI) was calculated using the equation described below (van Den Dool and Kratz 1963) using alkane standard solution C8-C20 (04070, Sigma-Aldrich, USA).

where n and n+1 are the number of carbons of standard alkane solution and t_x, t_n, and t_n+1 are retention times of a volatile compound of three wild Chrysanthemums (t_n < t_x < t_n+1).

To assess differences in the composition of volatile compounds of three wild Chrysanthemums through four flowering stages, the Venn diagram was generated using R version 4.1.0 (https://www.r-project.org/). Statistical analysis was performed using MetaboAnalyst5.0 (https://www.metaboanalyst.ca/home. xhtml, Pang et al. 2021). A data file of each Chrysanthemum was uploaded as CSV file format and filtered using interquartile range. Peak area was normalized by median values, transformed to generalized logarithm and then scaled by Pareto scaling (mean-centered and divided by the square root of the standard deviation of each variable). To compare the differences of volatile compounds among three Chrysanthemums, the data file was estimated by removing features with > 50% missing values and then estimated the remaining missing values using BPCA (Bayesian principal component analysis) method (Stacklies et al. 2007) because 30, 15, 24 out of 100 missing volatile compounds were found among C. boreale, C. indicum, and C. indicum var. albescens, respectively.

The one-way analysis of variance (ANOVA) with Tukey`s HSD (p < 0.05) for post-hoc analysis was used to identify the significance of volatile compounds among four flowering stages and three Chrysanthemum species (n = 3). Chemometric analysis was performed using PCA (principal component analysis) and PLS-DA (partial least squares-discriminant analysis). In addition, HCA (hierarchical clustering analysis) was used to generate heatmap. To generate heatmap, Euclidean distance was used to measure distance and Ward`s linkage method was used as clustering algorithm.

Results

Composition of volatile compounds in three chrysanthemums

Total 100 volatile compounds with > 90% of matching ratio form the library search were selected among three wild Chrysanthemums (Fig. 1). Total 70 volatile compounds were detected from C. boreale, and total 85 and 76 volatile compounds were detected from C. indicum and C. indicum var. albescens, respectively. In addition, 60 volatile compounds were overlapped between C. boreale and C. indicum and 67 compounds were overlapped between C. indicum and C. indicum var. albescens while 53 compounds were shared between C. boreale and C. indicum var. albescens implying C. indicum and C. indicum var. albescens have similar profile of volatile compounds. Furthermore, 49 volatile compounds were detected from all three wild Chrysanthemums and there were six, seven, and five compounds detected only in C. boreale, C. indicum, and C. indicum var. albescens, respectively. The six compounds only detected in C. boreale were β-thujene, trans-alloocimene, 1,3,8-p-Menthatriene, benzyl Alcohol, elixene, and benzophenone. On the other hand, β-pinene, 6-Methyl-5-heptene-2-one, 1,3,8-p-Menthatriene, α-longipinene, β-selinenol, α-eudesmol, and methyl palmitate were the seven compounds detected only in C. indicum. Finally, benzaldehyde, D-limonene, chavicol, β-linalool, and 1-perillaldehyde were the five compounds detected only in C. indicum var. albescens (Table 1).

70, 85, and 76 volatile compounds detected in C. boreale, C. indicum, and C. indicum var. albescens according to four different flowering stages were used to identify volatile compounds that show significance using One-way Analysis of Variance (ANOVA) by Tukey`s HSD (p < 0.05) as post-hoc analysis (Data not shown). 42 out of 70 volatile compounds in C. boreale showed significance at p < 0.05 differing their peak intensity during the flowering. 79 and 65 volatile compounds were found they showed significance at p < 0.05 in C. indicum and C. indicum var. albescens, respectively. In addition, when 100 volatile compounds detected from three wild Chrysanthemums were pooled together to compare the differences of volatile compounds composition of all three Chrysanthemums according to four flowering stages, 77 volatile compounds showed significance (p < 0.05) implying a variety of volatile compound would change the profile of volatile compounds during the flowering.

Analysis of temporal change of volatile compounds using multivariate analysis

In order to compare volatile compound changes among three different species according to four flowering stages, peak intensity of 42, 79, and 65 volatile compounds in each three wild Chrysanthemum was compared individually. In addition, all peaks were pooled and filtered as mentioned in Materials and Methods resulting in 82 volatile features useful for statistical analysis. PCA and PLS-DA were performed using 82 volatile features and, in the current study, PLS-DA was selected to explain differences of volatile compounds because PLS-DA showed stronger and clearer classification performance than the result of PCA.

To evaluate change of volatile compounds in three wild Chrysanthemums, the plots of PLS-DA were generated from volatile compounds of C. boreale, C. indicum, and C. indicum var. albescens according to different flowering stages. PLS-DA results showed clear differences between four different flowering stages of each species and when all three Chrysanthemums were compared together (Fig. 2). The four flowering stages based on peak intensity of volatile compounds of each species could be separated using two principal components (PC1 and PC2). In C. boreale, PC1 accounting for 45.8% showed the profile of volatile compounds differed between BF and other following stages (BO, HO, and FO stages) (Fig. 2A). On the other hand, flowering stage of C. indicum with 56.9% of PC1 was separated to three different groups; BF, BO and HO/FO (Fig. 2B). In both of C. boreale and C. indicum, three flowering stages could be completely separated by PC2 (18.7 and 29.4%, respectively). The profile of the volatiles of C. indicum var albescens with 73.1% of PC1 was separated to two groups (BF/BO and HO/FO, respectively) showing different grouping pattern unlike C. boreale and C. indicum (Fig. 2C). Two different stages in the same group were also clearly separated by PC2 (11.9%) suggesting the profile of volatile compounds in quantity would differ according to flowering stages. In addition, PLS-DA was performed to compare volatile compounds of all three wild Chrysanthemums together according to four different flowering stages. PC1 accounted for 40.3% and could separate three wild Chrysanthemums into three distinct groups (Fig. 2D). In this result, BO, HO, and FO stages of C. boreale showed a similar pattern and clustered together. In addition, three groups were generated by PC2 (17.4%). The volatiles of C. indicum at BF stage and two stages (HO and FO stages) of C. indicum var. albescens were separated as a different group while the other stages of three wild Chrysanthemums belonged to the similar group. 24, 28, and 32 volatile features with VIP score > 1.00 were found from C. boreale, C. indicum, and C. indicum var. albescens, respectively, and 26 volatile features from all volatile compounds were found in all three wild Chrysanthemums. The top ten volatile features are shown in Fig. 3.

The volatile compounds of three wild Chrysanthemums according to the four different flowering stages were compared using HCA (Fig. 4). Volatile compounds higher then mean value in heatmap is represented by the red color while the blue color accounts for the content of volatile compounds lower than corresponding mean value. 29 volatile compounds of C. boreale framed in the yellow box (frame A) in Fig. 4 were produced higher in C. indicum and C. indicum var. albescens. Those volatile compounds were eucalyptol, α-cadinene, α-phellandrene, α -gurjunene, 3-allylguaiacol, pivarose, ylangene, α-terpineol, caryophyllene oxide, (-)-α-cubebene, bornyl acetate, etc. Unlike 29 volatile compounds, 16 volatile compounds were produced less in C. indicum var. albescens than C. boreale and C. indicum (frame B). HCA could clearly separate the profile of volatiles compounds showing close relationship between C. indicum and C. indicum var. albescens other than C. boreale.

Discussion

Volatile compounds are a complex which is composed of a variety of small molecules produced from plants (Dudareva et al. 2013). More than 1,700 volatile compounds have been reported from > 900 plant species using head-space (HS) method (Knudsen et al. 2006). Volatile compounds are known to be spatially and temporally regulated. Flowers are the major organs that emits volatile compounds and leaves or stems also produce volatile compounds (Dötterl and Jürgens 2005;García et al. 2021). In addition, volatile compounds change depending on developmental stages of flowers, fruits, and leaves (Barman and Mitra 2019;Dudareva et al. 2000). However, only a couple of studies have been reported on the change of volatile compounds in Chrysanthemums. In C. boreale, growing environment such as fertilization can affect the yield of terpene and the mount of essential oil also changes between vegetative and reproductive stages (Kim et al. 2018; Lee et al. 2005).

The change of volatile compounds of three wild Chrysanthemums according to flowering stages were clarified and visualized using metabolomics approach with multivariate analysis and it was found that C. boreale showed different pattern of volatile compound changes rather than C. indicum and C. indicum var. albescens. Phenotypic evidence also showed three wild Chrysanthemums were grouped together implying closer relationship among them other than the other wild Chrysanthemum species (Kim et al. 2014). Phenotypic traits of C. indicum and C. indicum var. albescens are much similar than those of C. boreale and the same result was obtained from the current study when the composition of volatile compounds of three wild Chrysanthemums were compared.

Volatile compounds composition synthesized in plants are largely affected by physiological condition such as; the enzyme activity of a particular compounds, the available substrate, transcriptional regulation, and post-transcriptional regulation (Maeda and Dudareva 2012;Hemmerlin et al. 2012). Therefore, when other omics such as genomics, transcriptomics, or proteomics is combined with metabolomics approach, it will deepen our understanding of the mechanism and metabolism of volatile compounds produced in wild Chrysanthemums to improve volatile compounds both in quality and in quantity.

A comparative study of volatile compounds has been considered very useful to understand of surviving strategies in terms of adaptations and evolutionary processes at a level of species or subspecies (Knudson et al. 2006). Therefore, metabolomics approach using HS-SPME-GC-MS with multivariate analysis used in the current study will be a very powerful technique to evaluate temporal and spatial changes and differences of volatile compounds of floricultural plants. Furthermore, the current study will be useful not only to understand change the composition of the volatiles but also to choose the best condition of flowers to extract essential oils containing particular compounds for floricultural industry.