Integrated volatile metabolomics and transcriptomics reveal the regulation of odor components in celery exposed to supplementary blue light

Abstract

Light regulates the biosynthesis of plant volatile flavor metabolites, yet how supplemental blue light modulates the accumulation of volatiles in distinct celery (Apium graveolens L.) tissues and the associated molecular mechanisms remain largely unclear. In the present study, volatile metabolomics and transcriptomics approaches were employed to analyze volatile organic compounds (VOCs) and gene expression levels in celery leaves and petioles under supplementary blue light. The results revealed that supplementary blue light induced 251 and 318 differentially abundant metabolites in celery leaves and petioles, respectively. The content of 16 compounds (4 upregulated and 12 downregulated) in celery leaves significantly changed, including those with fruity, sweet, green, popcorn, and vegetable notes. Furthermore, the content of 42 compounds (31 upregulated and 11 downregulated) associated with sweet, green, fruity, cucumber, and citrus notes in celery petioles significantly changed. The combined transcriptomic and metabolomic analysis results revealed significant enrichment in phenylpropanoid metabolic pathway in celery petioles. Specifically, supplementary blue light significantly reduced CAD and CCR gene expression levels in celery, potentially reducing the contents of methyl chavicol and anethole. In conclusion, supplemental blue light significantly altered the accumulation profiles of flavor volatiles across different celery tissues, with metabolic changes concentrated in petioles and showing prominent tissue specificity. Moreover, blue light reprogrammed phenylpropanoid metabolism in petioles, driving the alterations of flavor-related volatiles. Our research clarifies the physiological basis of light-mediated flavor formation in vegetables and offers a foundation for elucidating the underlying metabolic pathways involved in plant secondary metabolism.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12870-026-09385-x.

Introduction

Celery (Apium graveolens L.), an important vegetable crop in the Apiaceae family, is widely cultivated in temperate and subtropical regions worldwide [1]. Celery is rich in dietary fiber, vitamins, and a variety of bioactive substances, including phenolic acids, flavonoids, and antioxidant polysaccharides [2, 3]. Therefore, celery is extensively utilized in cooking, food processing, and the medicinal field because of its nutritional value and sensory attributes.

The nutritional quality and sensory characteristics of vegetables may be influenced by a variety of environmental factors, such as light, temperature, and nutrition [2, 4, 5]. Among these environmental factors, light serves as the energy source for plant photosynthesis and selectively activates photoreceptors through light quality, thereby modulating gene expression and enzymatic activity to influence plant morphogenesis and secondary metabolite synthesis [6–8]. Blue light (400–499 nm), a key component of the visible spectrum, is uniquely positioned to concurrently drive photosynthetic activity and cryptochrome-mediated signaling, making it essential to elucidate its physiological and metabolic effects [9, 10]. Studies have demonstrated that blue light effectively increases the accumulation of bioactive compounds such as vitamin C, phenolic compounds, anthocyanins, and flavonoids in tomato (Solanum lycopersicum L.) [11], pepper (Capsicum annuum L.) [12], and cabbage (Brassica oleracea) [13]. For instance, continuous blue light exposure upregulates the expression of anthocyanin-related genes in pepper leaves, significantly increasing the accumulation of anthocyanins [12]. Additionally, blue light promoted the accumulation of phenolics, glucosinolates, and carotenoids in Chinese cabbage, thereby enhancing flavor profiles [13]. Nonetheless, the regulatory mechanisms underlying blue light-mediated flavor-related metabolism in leafy vegetables remain inadequately characterized.

Recent investigations have revealed that the distinctive flavor of celery primarily arises from its volatile organic compounds, which include terpenoids, esters, aldehydes, and alcohols [14]. These phytochemical constituents contribute to its distinctive aroma and are significantly correlated with antioxidant capacity and therapeutic potential [15, 16]. A total of approximately 30 characteristic terpenoids such as pinene and limonene have been identified in the main edible parts of celery (leaves and petioles) [17]. In addition, significant metabolic heterogeneity of VOCs was observed among different parts of plants [18]. For instance, the content of phenolic compounds in the pulp of citrus fruits is significantly lower than that in the peel [19]. The content of sesquiterpenes in the carpels of Magnolia laevifolia is lower than that in leaves and seeds, however, the content of oxygenated monoterpenes in carpels is greater than that in leaves and seeds for oxygenated monoterpenes [20]. Ingallina et al. [21] demonstrated that compared with petioles, celery leaves contain significantly higher levels of total phenolic acids, tannins, and flavonoids. Nevertheless, the synthesis of volatile compounds across different celery tissues under varying light environments, as well as the underlying mechanisms governing the distribution of light-influenced plant volatile flavors in various plant parts, remain to be further elucidated. Therefore, clarifying the main flavor compounds of different parts of celery is conducive to understanding the taste characteristics of celery as an edible vegetable and the processing applicability of different parts as a spice raw material.

Recent studies have shown that preharvest supplementation with blue light can enhance the production of beneficial flavor compounds, thereby improving the sensory quality of tomatoes and increasing consumer acceptance [22]. Nevertheless, how exposure to blue light during the plant growth period regulates flavor attributes in different parts of celery remains to be elucidated. In addition, gas chromatography‒mass spectrometry (GC‒MS), is a technique that combines the high-resolution separation capability of gas chromatography with the high-sensitivity detection and structural confirmation capability of mass spectrometry, making it an authoritative method for the qualitative and quantitative analysis of volatile trace compounds in complex plant matrices [23–27]. Our study aims to systematically investigated the effects of blue light on flavor metabolite profiles in celery leaves and petioles and elucidated the gene regulatory networks governing the characteristic aroma of crops via an integrated transcriptome and metabolome approach. By correlating supplementary blue light with perturbations in key volatile components, we further clarify the regulatory mechanisms underlying blue light-mediated volatile metabolism in celery and furnish empirical evidence for optimizing celery flavor quality through light environment modulation. This work also addresses the critical knowledge gap in the tissue-specific flavor formation of vegetables in response to specific light spectra, thereby offering novel insights into the complexity of plant flavor biosynthesis.

Materials and methods

Plant materials and treatment design

Celery (Apium graveolens L. cv. Dayehuang) seeds were procured from Qingdao Macaco Ecological Agriculture Co., Ltd, Shandong Province, China. The celery cultivar used in this study is a primary cultivar of Majiagou celery, a nationally certified geographical indication (GI) agricultural product in China. Celery seedlings exhibiting five true leaves were uniformly selected and transplanted into 1.1 L plastic pots (diameter, 13.5 cm, depth, 11 cm) in a plant factory with artificial lighting (PFAL). The pots contained a substrate mixture of peat, vermiculite, and perlite in a 3:1:1 (v/v/v) ratio. In this experiment, the white plus red LEDs (WR, Zhongshan Aier Lighting Technology Co., Ltd., Zhongshan City, China) treatment provided a photosynthetic photon flux density (PPFD) of 200 µmol·m⁻²·s^{− 1} and a photoperiod of 12 h d^{− 1} (WR). Supplementary blue light with light intensity of 20 µmol·m^{− 2}·s^{− 1} was added to the WR LEDs based on our previous study [28], which was designated as B20. The air temperature was maintained at 25 ± 1°C/18 ± 1°C (light/dark), and the relative humidity was maintained at 60–70% in the PFAL during experimental period. Celery plants were harvested at the commercial maturity stage, given that flavor enhancement is more consistent with consumer sensory preferences. To ensure valid and reliable experimental outcomes, block design was adopted to control spatial heterogeneity and eliminate systematic errors via statistical correction of block effects. In this study, each treatment was replicated in three blocks and 40 celery plants were grown within each block. At harvest, 5 uniform plants were randomly chosen from each block for measurement.

Sample preparation and extraction

The preparation and extraction of the leaves (L) and petiole (P) samples were completed by Wuhan Metware Biotechnology Co., Ltd. Uniformly sized, randomly selected fully expanded mature leaves and petioles were sampled at the harvest stage. Each treatment was subjected to three replicates and each replicate was derived from samples taken from five different celery plants randomly selected from the treatments. The sampling process strictly adheres to the principles of randomness, representativeness and consistency. Systematic errors are eliminated through the initial division, and random errors are reduced through the repetition setup to ensure the robustness of the data. The celery leaves and petiole samples from the various treatments were preserved by freezing at − 80°C to facilitate subsequent analyses. Samples were ground into a powder using liquid nitrogen. Subsequently, 500 mg of the powder was placed into a headspace vial along with a saturated NaCl solution and 20 µL (10 µg/mL) of the internal standard (3-Hexanone-2,2,4,4-d4). The 3-Hexanone-2,2,4,4-d4 (CDN Isotopes, Canada) served as the internal standard for quality control and semi-quantification [29, 30]. Automated headspace solid-phase microextraction (HS-SPME) was then used for extraction.

Gas chromatography-mass spectrometry conditions

For headspace solid-phase microextraction analysis, each vial was placed in 60°C for 300 s, then a 120 μm DVB/ CWR/PDMS fiber was exposed to the headspace for 15 min. Desorption of the volatile com pounds from the fiber coating was performed in the injection port of GC–MS (8890-7000D, Agilent, Santa Clara, CA, USA) at 250°C for 5 min in splitless mode. The chromatographic column was DB-5MS (30 m × 0.25 mm × 0.25 μm, Agilent), and the carrier gas was helium at a flow rate of 1.2 mL/min, with an inlet temperature of 250°C with a solvent delay of 3.5 min. The column temperature program was as follows: 40°C (3.5 min) ramping up to 180°C (3 min) at 3°C/min, and then ramping up to 250°C (5 min) at 6°C/min The temperature was then increased to 250°C (5 min) at 6°C/min. Operating in EI mode (70 eV), the MS featured an ion source temperature of 230°C, quadrupole at 150°C, and interface at 280°C based on the previous study [31, 32].

Identification and quantification of volatile metabolites

Scans covered m/z 35–350, with SIM employed for both qualitative and quantitative analysis. Qualitative was performed by NIST library comparison and retention index, and quantitative was based on peak area normalization method [33, 34]. During data preprocessing, missing values were first imputed as 1/5 of the minimum value corresponding to each row (metabolite). Next, the coefficient of variation (CV) was computed for the quality control (QC) samples; only metabolites with a CV less than 0.5 were retained to construct the final data file. The quantitative analysis of the volatile metabolome is carried out by comparing the chromatographic peak area of the analyte volatile substances with that of the internal standard (3-hexanone-2,2,4,4-d4 as internal standard), and calculating the content of the volatile substances relative to the internal standard based on previous studies [35–38].

Screening of differential accumulated metabolites

Differential volatile metabolites in celery tissues were identified by: VIP value >1 and fold change ≥ 1.5 [39, 40]. Data preprocessing included log2 transformation and mean-centering. Orthogonal projection to latent structures-discriminant analysis (OPLS-DA) modeling then separated group-specific variation from orthogonal noise to screen discriminative compounds, implemented via Metabo AnalystR package in R software [41]. Metabolite annotation was conducted using the KEGG Compound database (https://www.kegg.jp/kegg/compound/), and the annotated metabolites were mapped to metabolic pathways using the KEGG Pathway database (https://www.kegg.jp/kegg/pathway.html). Network mapping linking sensory flavor characteristics and differential metabolites was established according to the inherent flavor-related properties of corresponding metabolites using Metware Cloud online analysis platform (https://cloud.metware.cn/, Metware Biotechnology Co., Ltd., Wuhan, China).

Calculation of relative odor activity value

To elucidate the contribution of each VOC to the overall aroma characteristics, the relative odor activity value (rOAV) approach was utilized to identify the key flavor components in the samples based on the previous studies [42, 43]. rOAV was calculated with equation: rOAV = Ci/Ti, where Ci and Ti represent the relative concentration and odor threshold in the water of the compounds.

RNA-seq and data processing

Total RNA was extracted from celery samples at harvest using the FastPure Plant Total RNA Isolation Kit (Vazyme, China) following the manufacturer’s instructions. RNA quantity and quality were examined using a Qubit 4.0 fluorometer and Qsep400 bio-analyzer. mRNA was enriched via Oligo (dT) magnetic beads and fragmented. First- and second-strand cDNA were synthesized sequentially, then purified. After end repair, dA-tailing and adapter ligation, cDNA fragments were size-selected and amplified by PCR to construct the library. The library was quantified and sized prior to Illumina sequencing. Then, the constructed RNA-seq library was sequenced, which was conducted by Wuhan MetWare Biotechnology Co., LTD. We performed rigorous quality control on the raw reads prior to data analysis. Specifically, we filtered out reads with adapter sequences, reads containing N bases accounting for over 10% of the total base number, and reads where bases with a Phred quality score ≤ 20 made up more than 50% of the total length. Subsequently, we assessed the sequencing error rate and GC content. Quantitative analysis of gene expression levels and differential analysis were based on previous studies [44]. The clean reads were aligned to the reference genome [45]. Clean reads were aligned to the reference genome using HISAT2 to obtain their positional information on the reference genome and genes, as well as unique sequence characteristics of the sequencing samples [46].

Transcriptome analysis

All mapped transcripts were analyzed for differentially expressed genes (DEGs) using the R package ‘edgeR’. Gene expression levels were calculated and normalized based on FPKM values. The false discovery rate (FDR) was applied to set the threshold for P-values across multiple tests. The screening conditions for DEGs are: |log2FoldChange| ≥ 1, and FDR < 0.05. KEGG pathway enrichment was evaluated using the rich factor, q-value, and the count of enriched genes per pathway based on previous studies [47]. Pearson correlation analysis was performed after log2 transformation of the normalized datasets. Gene–metabolite pairs with correlation coefficients (|r| > 0.8) and P-values < 0.05 were retained to construct the gene–metabolite correlation network. Integrated KEGG enrichment analysis was performed on differentially expressed genes and differential metabolites to identify co-enriched signaling pathways. The enriched pathways were visualized using Metware Cloud online analysis platform (https://cloud.metware.cn/, Metware Biotechnology Co., Ltd., Wuhan, China).

RT-qPCR analysis

Total RNA was extracted and reverse-transcribed into cDNA. cDNA levels were then measured by real-time quantitative PCR (RT-qPCR) with a commercial kit (Vazyme, Nanjing, China) according to the manufacturer’s instructions. The RT-qPCR operation was conducted on a Light Cycler 480 II (Roche, Germany). The PCR conditions were as follows: denaturation at 95°C for 5 min; 40 cycles of 95°C for 15 s, and 60°C for 30s. The mRNA levels of the celery reference gene were used to normalize the data for each RT-qPCR run. The relative gene expression was calculated based on the 2^−ΔΔCt method [48]. The RT-qPCR primer sequences are shown in supplementary file: Table S1 and the reference gene sequence was used based on the previous study [49].

Statistical analysis

Statistical differences were determined by a Student’s t-test using SPSS 26.0 software (IBM, Inc., Chicago, IL, USA) (p < 0.05).

Results

Volatile metabolites following supplementary blue light

A total of 1389 volatile components of celery, including terpenoids, esters, heterocyclic compounds, and ketones, were identified. In addition, terpenoids accounted for 22.32% of the identified metabolites, followed by esters (16.49%), and ketones (10.8%) (Fig. 1a). Principal component analysis (PCA) showed significant differences between the WR and B20 samples, which could be categorized into four different groups (Fig. 1b). The results obtained using hierarchical clustering analysis, as visualized in the heatmap, were highly consistent with the PCA results (Fig. 1c). Blue light radiation induced 251 differentially abundant metabolites (115 up-regulated and 136 down-regulated) in celery leaves and 318 differentially abundant metabolites (212 up-regulated and 106 down-regulated) in celery petioles (Fig. 1d). A total of 102 differentially abundant metabolites were coregulated in leaves and petioles and were classified into 11 categories, including terpenoids, esters, alcohols, ketones, heterocyclic compounds, acids, amines, aldehydes, hydrocarbons, ethers, and phenols (Fig. 1e).

Fig. 1.

Volatile metabolite analysis. a Classification and composition. b Principal component analysis (PCA) of samples. c Hierarchical clustering of metabolites in different celery parts under blue light. d Counts of differentially accumulated metabolites (DAMs). e Venn diagram of DAMs in leaves (L) and petioles (P) of celery under blue light (B20-L, B20-P) versus untreated controls (WR-L, WR-P)

Multivariate statistical analysis of differential accumulate metabolites

Volatile compounds in celery leaves and petioles exposed to blue light radiation were clearly discriminated from those under WR (Fig. 2a-b). This distinct separation underscores that blue light radiation induces a comprehensive reprogramming of volatile metabolism, leading to metabolomes that are qualitatively distinct from those grown under control. Validation of the OPLS-DA model through substitution testing confirms that the model is not overfitted, indicating that the model has good explanatory and predictive capabilities for further analysis (Fig. 2c-d). Volcano analysis revealed 69 upregulated and 98 downregulated volatile metabolites in blue-light-treated celery leaves relative to the control, whereas 182 metabolites were significantly induced and 100 metabolites were repressed in petiole tissue (Fig. 2e-f).

Fig. 2.

Screening and classification of differentially accumulated metabolites (DAMs) in celery under blue light. a-b OPLS-DA score plots comparing B20-L vs. WR-L (leaves) and B20-P vs. WR-P (petioles). c-d Permutation test (n=200) of the OPLS-DA model comparing metabolite profiles in celery leaves and petioles under different treatments. e-f Volcano plots illustrating differentially accumulated volatile metabolites from comparisons of B20-L vs. WR-L and B20-P vs. WR-P

K-means analysis revealed four characteristic clusters of metabolite fluctuations in celery leaves versus petioles under blue light exposure (Fig. 3). Subclasses 4 showed a significant increase in volatile metabolite content after blue light irradiation. The more pronounced accumulation of these metabolites in celery leaves implies a tissue-specific strategy.

Fig. 3.

K-means clustering analysis of differentially accumulated metabolites of different parts of celery exposed to blue light. The y-axis shows the standardized amount of each metabolite, and the x-axis shows the different samples. B20-P and B20-L represent the samples of celery petioles and leaves exposed to blue light, respectively. WR-P and WR-L represent the samples of celery petioles and leaves not exposed to blue light, respectively

KEGG enrichment analysis of differential accumulate metabolites

KEGG database was used to annotate and enrich the differential metabolites between B20 treatment and WR treatment. Our results suggest that blue light may enhance the provision of amino acid precursors and energy in leaves, potentially supporting the accelerated synthesis of flavor and defense-related volatile compounds. Conversely, in celery petioles, the enrichment was primarily in phenylpropanoid biosynthesis and glutathione metabolism. This induction of defense-related pathways indicates that petioles may activate a more robust antioxidant and defense system under blue light stress, possibly involving the production of protective phenolic compounds and the scavenging of reactive oxygen species (Fig. 4a-b).

Fig. 4.

Kyoto Encyclopaedia of Genes and Genomes (KEGG) enrichment analysis of differentially accumulated metabolites (DAMs) and in-depth analysis of key metabolites. a KEGG enrichment analysis for DAMs in B20-L and WR-L. b KEGG enrichment analysis for DAMs in B20-P and WR-P. Rich factor is defined as the ratio of the number of differential metabolites in a given pathway to the total number of metabolites annotated to this pathway; a higher value indicates greater enrichment. A smaller P-value denotes more significant enrichment. The size of each dot in the plot corresponds to the number of significantly differential metabolites enriched in the respective pathway. WR-P and WR-L represent the samples of celery petioles and leaves not exposed to blue light. B20-P and B20-L represent the samples of celery petioles and leaves exposed to blue light

rOAV analysis and odor correlation analysis

The differential accumulation metabolites that contributed directly in the flavor of celery were obtained (Table S2; Table S4). The results showed that the odors directly produced by different metabolites in celery leaves after exposing to supplemental blue light radiation were mainly reflected in wood, fruity, herbal, sweet, green, waxy, spicy, fresh, balsamic, floral (Fig. 5a), the odors directly contributed by different metabolites in celery petioles were mainly embodied in green, sweet, nutty, fresh, waxy, apple, herbal, floral, fruity, woody (Fig. 5b). Supplementary blue radiation inhibited green and sweet odors in celery leaves. Under the influence of different compounds, the various aromatic components (including herbal, fruity, woody, floral, balsamic, fresh, spicy, and waxy notes) exhibited a general downward trend, albeit with distinct regulatory patterns (Fig. 5c and e). Supplementary blue radiation can promoted waxy odors in celery petioles, while scents such as sweet, green, woody, fruity, floral, herbal, fresh and nutty can both promote and inhibit (Fig. 5d and f).

Fig. 5.

Flavor omics analysis of volatile differential accumulated metabolites. a Flavor wheel of differentially accumulated metabolites (DAMs) in group B20-L vs. WR-L. b Flavor wheel of DAMs in group B20-P vs. WR-P. c Sankey diagrams between aroma and differential volatile metabolites in group B20-L vs. WR-L. d Sankey diagrams between aroma and differential volatile metabolites in group B20-P vs. WR-P. e Network diagram between aroma and differential volatile metabolites in B20-L vs. WR-L. f Network diagram between aroma and differential volatile metabolites in B20-P vs. WR-P. B20-L and B20-P represent the samples of celery leaves and petioles exposed to blue light. WR-L and WR-P represent the samples of celery leaves and petioles not exposed to blue light

Differential accumulate metabolites on the celery flavor

Metabolomic profiling demonstrated that supplementary blue light significantly modulates the accumulation patterns of flavor related metabolites in celery leaves and petioles (Fig. 6a). The most relevant differential metabolites for flavor were concentrated on seven compounds (Fig. 6b). Specifically, acid, ester, ketone, and aldehyde were up-regulated by 1, 1, 1, and 1 substance, respectively; While esters, alcohols, ketones, heterocyclic compounds and terpenoids were down-regulated by 5, 1, 3, 2 and 1 substances, respectively (Fig. 6c). The most flavor-related differential metabolites in celery petioles were concentrated on nine compounds. Esters, heterocyclic compounds, alcohols, ketones, terpenoids, ethers, aldehydes, and phenols were up-regulated by 4, 4, 1, 3, 8, 2, 6, and 3 substances, respectively. Esters, alcohols, terpenoids, aromatics and aldehydes were down-regulated by 3, 1, 4, 1 and 2 substances, respectively (Fig. 6d).

Fig. 6.

Differentially accumulated metabolites (DAMs) in key flavor compounds. a Heat map comparing B20-L and WR-L. b Heat map comparing B20-P and WR-P. c Classification and quantity of DAMs in B20-L and WR-L. d Classification and quantity of DAMs in B20-P and WR-P. Red indicates a relatively high metabolite abundance, and green indicates a relatively low abundance. B20-L and B20-P represent the samples of celery leaves and petioles exposed to blue light. WR-L and WR-P represent the samples of celery leaves and petioles not exposed to blue light

Global transcriptional changes in response to supplementary blue light

A total of 94.9 Gb of clean data was obtained in our study. The percentages of Q30 and GC were 92.76–93.95% and 45.06–46.69%, respectively. The alignment efficiency against the reference genome Apium_graveolens.genome.fasta was 94.43–95.71%, indicating a relatively high performance (Table S3). Principal component analysis of each group indicated significant differences between WR and B20 samples, which can be divided into four groups based on inter-sample distances (Fig. 7). Comparison of gene expression levels between B20 treatment and WR treatment. The results revealed that 427 and 736 genes were up-regulated and 1680 and 1819 genes were down-regulated in leaves and petioles of celery, respectively (Fig. 7). KEGG pathway analysis showed that the significant metabolic pathways and signaling pathways of differentially expressed genes (DEGs) in celery leaves exposed to blue light treatment were plant-pathogen interactions (124 DEGs), MAPK signaling (93 DEGs), biosynthesis of unsaturated fatty acids (25 DEGs) and phenylpropanoid biosynthesis (29 DEGs), while those of differentially expressed genes in celery petioles were the MAPK signaling (78 DEGs), plant-pathogen interactions (104 DEGs) and photosynthesis-antenna proteins (9 DEGs) (Fig. 7).

Fig. 7.

Screening and classification of differentially expressed genes (DEGs) in celery exposed to blue light. a PCA score plots with each point representing an independent biological replicate. b-c Hierarchical cluster analysis of DEGs in different parts of celery after exposure to blue light between B20-L vs. WR-L, B20-P vs. WR-P, respectively. d-e Volcano plot of DEGs between B20-L vs. WR-L, B20-P vs. WR-P, respectively. f-g Kyoto Encyclopaedia of Genes and Genomes (KEGG) enrichment analysis of DEGs and in-depth analysis of key genes between B20-L vs. WR-L, B20-P vs. WR-P, respectively. The rich factor is defined as the ratio of DEGs enriched in a specific pathway to all genes annotated to the corresponding pathway. A higher rich factor corresponds to stronger enrichment, while a smaller Q-value denotes more statistically significant enrichment. B20-L and B20-P represent the samples of celery leaves and petioles exposed to blue light. WR-L and WR-P represent the samples of celery leaves and petioles not exposed to blue light

Integrated analyses of metabolomics and transcriptomics in celery petioles in response to supplementary blue light

The results of KEGG enrichment analysis involving differentially expressed genes (DEGs) and differentially accumulated metabolites (DAMs), secondary metabolite biosynthesis, phenylpropanoid biosynthesis, and other pathways showed significant enrichment in both DEGs and DAMs, suggesting that these pathways may be significantly activated or regulated under blue light treatment (Fig. 8). This coordinated response at the transcriptional and metabolic levels suggests that these pathways are core targets of blue light signaling, likely activated as part of a generalized defense or acclimation response. Up-regulation of the expression of genes involved in most of the key steps in celery after blue light irradiation promoted the metabolism of the phenylpropanoid biosynthetic pathway and accelerated the production of related compounds (Fig. 8b). The expression of 6 genes randomly selected in the phenylpropane metabolic pathway in the two treatments of celery leaves was detected by RT-qPCR (Fig. 8c). The transcriptome sequencing data were validated by RT-qPCR, and six differentially expressed genes (Ag10G01955, Ag8G01103, Ag9G01199, Ag8G00380, Ag3G02794, and Ag1G01471) were selected and analyzed for their relative expression under blue light treatment. The correlation between RNA-seq (Fig. 8b) and RT-qPCR data (Fig. 8c) confirms the robustness of our transcriptional profiling and solidifies these genes as reliable candidates for future functional studies on blue light-mediated regulation.

Fig. 8.

a Kyoto Encyclopaedia of Genes and Genomes (KEGG) enrichment analysis of differentially expressed genes (DEGs), differentially accumulated metabolites (DAMs) and in-depth analysis of key genes and key metabolites between B20-P vs. WR-P. b The DEGs between B20-P vs. WR-P in the KEGG pathway of phenylpropanoid biosynthesis. c RT-qPCR validation of the transcriptome data results for 6 selected genes between B20-P vs. WR-P. Bars and dots indicate gene expression levels and metabolite contents, respectively. PAL, phenylalanine ammonialyase; C4H, cinnamate-4-hydroxylase; 4CL, 4-coumarate-CoA ligase; CCR, cinnamoyl-CoA reductase; CAD, cinnamyl alcohol dehydrogenase. d Network diagram between aroma components and sensory characteristics between B20-P vs. WR-P. e Correlation network diagram between compounds and genes between B20-P vs. WR-P. B20-P represents the samples of celery petioles exposed to blue light. WR-P represents the samples of celery petioles not exposed to blue light

Discussion

As one of the critical environmental factors, light can influence plant growth and regulate the accumulation and metabolism of volatile compounds, thereby altering plant characteristics and nutritional quality. Therefore, determining appropriate light conditions is essential for formulating strategies to enhance crop performance [50–53]. Blue light plays a pivotal role in regulating metabolic responses in plants [54]. For example, blue light significantly altered the content of photosynthetic pigments (e.g., carotenoids) in microgreens and affected the methylerythritol 4-phosphate (MEP) metabolic pathway [55]. Research has shown that blue light enhances cellular energy and redirects metabolic pathways in citrus fruits, which enables the synthesis of highly diverse natural products during primary and secondary metabolism [56]. Amino acids, alkaloids, and intricate phenylpropanoid and isoprenoid derivatives, are among these metabolites, with the latter two categories encompassing carotenoids and terpenoids [56].

To investigate the effects of blue light on volatile flavor compounds in celery, we analyzed the samples using headspace solid-phase microextraction coupled with gas chromatography–mass spectrometry (HS-SPME-GC-MS). This approach integrates the efficient and non-destructive enrichment of volatile compounds via headspace solid-phase microextraction with the exceptional separation, qualitative, and quantitative capabilities of GC-MS, making it a robust analytical platform for in-depth characterization of complex aromatic profiles in plants [27, 57–60]. Our results indicated that a total of 1,389 volatile compounds were identified in this study, with terpenoids, esters, and ketones constituting the major components, and terpenes constituting approximately 22.32% of the overall compounds detected (Fig. 1). Our results confirm that terpenoids are the most predominant VOCs in celery, which is similar to previous findings [14, 60]. In addition, previous studies revealed that volatile compounds in different organs of Houttuynia cordata Thunb [61] and Camellia sinensis (L.) O. Kuntze [62] exhibited different responses to blue light irradiation. Similarly, our results indicated that the volatile compounds in different parts of celery respond differently to supplementary blue light. Supplemental blue light was found to significantly enhance the accumulation of terpenoids and aldehydes in celery petioles at the mature stage, demonstrating that supplemental blue light radiation significantly altered the formation of flavor compounds in celery.

Celery has a distinct aroma, and its volatile organic compounds have been extensively studied [14, 17, 63]. The primary constituents of celery include a series of terpenoids, phenolic compounds, ester compounds, and aromatic compounds [64, 65]. Volatile compounds contribute to aroma formation. Among these volatiles, volatile terpenoids are predominantly synthesized via the methylerythritol phosphate (MEP) pathway catalyzed by terpene synthases (TPS). The differences in aroma among the celery in different treatments of celery resulted from variations in volatile components, which may be attributed to either the presence of specific compounds or the proportional composition of all the aroma contributors [66]. Studies have shown that terpenes, phthalides, and aldehydes are the organic compounds responsible for the characteristic aroma and flavor of celery [17]. The volatile compound profiles in celery petioles and leaves are relatively similar, collectively contributing to its characteristic herbaceous and green aroma notes. However, the specific sensory profile can vary significantly due to differences in chemical composition, which are influenced by multiple factors including genotype, season, plant part consumed, geographical region of growth, harvest stage, and post-harvest handling [67]. While much of the existing literature focuses on the levels of terpenes and aromatic esters, other compounds present in celery, such as alcohols, esters, and aldehyde, should not be overlooked, as they play a crucial role in imparting fresh, grassy, and green aroma nuances [67]. Our results indicate that terpenes, esters, aldehydes, and phenolic compounds in celery petioles treated with supplementary blue light significantly increased, contributing to the enhancement of flavors such as green and woody notes (Fig. 5). Moreover, previous research findings indicate that γ-terpinene, d-limonene, (E)-2-hexenal, and β-myrcene contribute contributed primarily to the aroma of celery [14]. Furthermore, the differences in aroma between pomelo (Citrus maxima L.) and sweet orange (Citrus sinensis L.) are primarily attributed to variations in specific volatile compounds, such as valencene and nootkatone [68].

Supplementary blue light also caused differences in the characteristic volatile compounds of celery. Our results revealed that the contents of (E)-2-hexenal, 2-hexenal, 2-propenoic acid butyl ester, (1S)-2,6,6-trimethylbicyclo[3.1.1]hept-2-ene, (1R)-2,6,6-trimethylbicyclo[3.1.1]hept-2-ene, and anethole in celery petioles significantly increased after supplementary blue light, with primary flavor characteristics described as fruity, cucumber, citrus, and green (Figs. 5 and 6). Supplementary blue light effectively modulates the composition of characteristic volatile compounds in celery, thereby increasing the overall accumulation of fruity and fresh flavor substances and altering the flavor quality of celery. In addition, Our distinct regulatory results indicate a tissue-specific strategy: celery petioles undergo a comprehensive up-regulation of diverse flavor volatiles, likely to directly enhance aroma and taste. In contrast, celery leaves exhibit a down-regulation trend in blue light radiation, potentially redirecting metabolic resources towards other light-responsive processes such as defense or photosynthesis. In addition, blue light induced greater variations in the abundance of volatile flavor constituents in celery petioles relative to leaves and key flavor compounds were mainly upregulated in petioles while suppressed in leaves (Figs. 2 and 6), confirming the existence of tissue specificity. Analogously, earlier studies have documented that volatile metabolomes differed considerably among various white tea tissues: terpenoids were the predominant metabolites accumulated in tender stalks, leaves chiefly produced volatile derivatives originating from fatty acids, while apical buds contained evenly distributed phenylpropanoid metabolites [69].

Transcriptome analysis revealed that supplementary blue light treatment induced differential expression of numerous genes. Specifically, a total of 2,555 differentially expressed genes were detected in the celery petioles, whereas only 2,107 differentially expressed genes were detected in the celery leaves exposed to supplementary blue light. These differentially expressed genes are associated with phenylpropanoid biosynthesis, glutathione metabolism, and fatty acid degradation pathways. Integrated transcriptomic and metabolomic analyses indicated that the impact of supplemental blue light irradiation on celery flavor is mediated primarily through the phenylpropanoid metabolic pathway (Fig. 7). Multiomics integration revealed that supplementary blue light modulates celery flavor formation primarily through tissue-specific regulation of the phenylpropanoid biosynthesis pathway.

Phenylpropanoids are highly diverse specialized metabolites that originate from the shikimate pathway and the phenylpropanoid pathway and generate large amounts of lignin and flavonoid-related metabolites, which are ubiquitous throughout the plant kingdom [70]. These compounds play critical roles in plant responses to environmental challenges. Plants utilize shikimate-derived precursors to generate a vast array of functionally essential specialized metabolites via this metabolic pathway [71]. Different light qualities can significantly affect the synthesis and accumulation of phenylpropanoid compounds in plants. For example, the content of volatile phenylpropanoids in sweet basil (Ocimum basilicum) leaves is regulated by the red/far-red light ratio; an increase in this ratio elevates the concentrations of phenolic acids (predominantly phenylpropanoids) such as rosmarinic acid and cichoric acid in the leaves [72]. These findings suggest that adjusting the light spectrum can specifically modulate the production of certain phenylpropanoid flavor compounds. In our study, celery regulated by blue light resulted in significant alterations in the flavor compounds within the phenylpropanoid metabolic pathway. These findings indicate that supplementation with narrow-band blue and red LED lighting can influences key aromatic volatiles in basil [73]. These results further confirm that light quality is a critical factor in regulating the formation of phenylpropanoid flavor compounds in plants. This implies that within plant factories or greenhouses, the approach can transcend the simplistic mindset of yield response to supplemental lighting and enter a stage of precise, quality-oriented management tailored to specific needs. Targeted enrichment of specific flavor compounds can be achieved, yielding customized premium celery products with stable, intense characteristic flavors to meet the demands of specific consumer markets.

Isoeugenol, a key phenylpropanoid-derived secondary metabolite in plants, is biosynthesized via the phenylpropanoid pathway and plays significant roles in defense and aroma [74]. The phenylpropanoid pathway thereafter branches into a complex network, with metabolic flux being finely regulated at multiple enzymatic checkpoints to determine the fate of shared precursors. Phenylalanine ammonia-lyase (PAL) plays a critical role in isoeugenol biosynthesis by catalyzing the deamination of phenylalanine to cinnamic acid [75]. The coordinated actions of 4-coumarate: CoA ligase (4CL) and cinnamyl alcohol dehydrogenase (CAD) constitute rate-limiting checkpoints in phenylpropanoid flux partitioning, directing metabolic precursors toward specialized metabolite biosynthesis. For instance, in Arabidopsis thaliana, a quadruple mutant of PAL genes (pal1/pal2/pal3/pal4) results in reduced accumulation of phenylpropanoid-derived lignin [76]. Isoeugenol biosynthesis shares common precursors and metabolic flux with lignin biosynthesis through the phenylpropanoid pathway. The observed variation in isoeugenol content may be attributed to altered activities of PAL and CAD enzymes governing phenylpropanoid flux partitioning [77].

Our results indicate that the expression of genes related to blue light radiation modulated pathways including phenylpropanoid biosynthesis and eugenol content significantly increased after supplementary blue light treatment, whereas methyl chavicol and anethole contents significantly decreased following blue light exposure (Fig. 8b). Specifically, significant correlations were observed between the accumulation of the differentially abundant metabolite isoeugenol and the expression level of the PAL genes. The expression levels of most genes detected by RT-qPCR were consistent with the results of the transcriptome analysis following blue light treatment. Studies have shown that under 4-HBA stress, CAD gene expression in the lignin synthesis pathway is significantly upregulated and can positively regulate the expression of metabolites in this pathway [78]. In our study, CAD genes positively regulated the levels of methyl chavicol and anethole. In addition, the biosynthesis of eugenol may be negatively regulated by CAD and CCR genes (Fig. 8). Our results indicate that within the phenylpropanoid metabolism pathway of celery, the blue light-regulated compound isoeugenol contributes clove and floral notes, while methyl chavicol and anethole impart sweet, stewed, flowery, and exotic flavors (Fig. 8d). Correlation analysis revealed that the genes Ag8G01103 and Ag9G01199 were significantly correlated with isoeugenol (Fig. 8e). However, the regulatory relationship between gene expression levels and the final accumulation of metabolites still requires further validation at the levels of protein activity and the flux of metabolic intermediates. Future studies could integrate metabolomics and proteomics to systematically analyze the posttranslational modification states of key enzyme proteins under blue light and their associations with dynamic changes in metabolites. Furthermore, the precursor competition between different metabolic branches also suggests that future research needs to further elucidate the fine regulatory mechanisms of these key enzymes at the levels of spatiotemporal expression, posttranslational modifications, and protein‒protein interactions. Supplemental blue light exhibits promising practical application prospects for celery cultivation under protected horticulture. Supplemental blue irradiation significantly elevates the concentrations of predominant volatile constituents responsible for the characteristic flavor and aroma of celery. Therefore, optimized blue light supplementation serves as a feasible physical regulating strategy to enhance the commercial flavor quality of celery in practical greenhouse production.

Conclusion

After supplemental blue light irradiation, a total of 16 and 42 significantly differential volatile metabolites were identified in celery leaves and petioles, respectively. Supplementary blue light reshaped the fruity, sweet, green and fatty sensory characteristics of both tissues, while cucumber, citrus and woody flavor profiles were specifically altered in petioles. In addition, the supplementary blue light markedly downregulated CAD and CCR gene expression levels, and suppressed CCR and CAD enzyme activity in the phenylpropanoid metabolic pathway of celery petioles, potentially reducing the content of methyl chavicol and anethole compounds. Our study reveals supplementary blue-light-induced VOC alterations across celery tissues and the gene network underlying petiole flavor metabolism, laying a theoretical basis for precise regulation of beneficial secondary metabolites via spectral management in controlled agriculture.

Authors’ contributions

Zhengnan Yan: Writing – review & editing, Methodology, Conceptualization, Supervision, Visualization, Funding acquisition. Zhixin Li: Writing–original draft, Methodology, Data curation, Software, Visualization, Investigation, Formal analysis. Chao Liu: Investigation, Data curation, Software. Ning Yang: Methodology, Supervision. Jinxiu Song: Methodology, Supervision. Na Lu: Methodology, Supervision. Kefeng Pu: Methodology, Supervision. Duo Lin: Methodology, Supervision. Yanjie Yang: Writing – review & editing, Conceptualization, Methodology, Supervision, Funding acquisition. All authors reviewed the manuscript.

Funding

This work was financially supported by the Modern Agricultural Industrial Technology System of Shandong Province (SDAIT-05), the Key Research and Development Program of Shandong Province (2024TZXD019), University-Enterprise Cooperation Project (6602423221), the Postgraduate Innovation Program of Qingdao Agricultural University (QNYCX24031), Innovation & Entrepreneurship Training Program for College Students of Qingdao Agricultural University (QNDC20250148 & QNDC20250164).

Data availability

The datasets generated and/or analysed during the current study are available in the National Center for Biotechnology Information (NCBI) with BioProject accession numbers PRJNA1427769.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.