Skip to main content
NCBI home page
Food Science & Nutrition logoLink to Food Science & Nutrition
. 2026 Jul 16;14(7):e72115. doi: 10.1002/fsn3.72115

Comparison of the Metabolites in Fingered Citron Fruit (Citrus medica L. var. sarcodactylis Swingle) and Chayote (Sechium edule) Based on UPLC‐Q‐Orbitrap MS/MS

Bin Li 1,2, Ruiyi Fan 1,2,✉
PMCID: PMC13376348  PMID: 42491200

ABSTRACT

Fingered citron ( Citrus medica L. var. sarcodactylis Swingle) is a medicinal and edible citrus fruit cultivated in several major production regions of China. However, comprehensive information on its regional metabolite variation and its chemical differentiation from chayote ( Sechium edule ), a botanically unrelated commodity that may be confused with fingered citron because of partially overlapping Chinese vernacular names, remains limited. In this study, untargeted UPLC‐Q‐Orbitrap MS/MS was used to profile fingered citron samples from Zhejiang (CF1), Sichuan (CF2), Guangdong/Guangxi (CF3), and Yunnan (CF4), together with chayote samples used as a targeted authentication comparator (CFM). A total of 380 metabolites were putatively annotated, including 47 secondary metabolites comprising 23 flavonoids, 8 terpenoids, 8 phenols, 3 coumarins, 3 alkaloids, and 2 steroids. Multivariate analyses and hierarchical clustering identified 65 differential metabolites, including 19 secondary metabolites, that clearly separated the five sample groups within the present dataset. The very high cross‐validated Q 2 values should nevertheless be interpreted cautiously because of the large taxonomic and metabolic distance between C. medica and S. edule and the limited sample size. The interspecific separation was therefore interpreted as an expected chemotaxonomic difference with practical authentication value, rather than as a comparison between biologically equivalent taxa. Phenylalanine metabolism and flavone and flavonol biosynthesis showed the strongest nominal enrichment, although neither remained significant after FDR correction. These results provide a metabolite reference for regional quality discrimination of fingered citron and identify candidate chemical features for its targeted authentication against chayote. Further validation using independent samples, additional production years, and other potential comparator commodities is required before routine application.

Keywords: chayote, Citrus medica , food authentication, regional quality discrimination, untargeted metabolomics

Untargeted UPLC‐Q‐Orbitrap MS/MS metabolomics was used to profile fingered citron ( Citrus medica ) from four Chinese production regions (CF1–CF4) and chayote ( Sechium edule , CFM). Of 380 putatively annotated metabolites, 65 differential metabolites distinguished the five groups, clearly separating chayote from fingered citron and highlighting regional chemical variation among fingered citron cultivars.

graphic file with name FSN3-14-e72115-g005.jpg

1. Introduction

Fingered citron (CF) ( Citrus medica L. var. sarcodactylis Swingle) also known as “Foshou” in China, is one member of the three ancestral species of Citrus genus (Mahdi, Al‐Ansi, Al‐Maqtari, et al. 2019). About 300 years ago, CF was merely cultivated as ornamental plant (Sottile et al. 2019). Nowadays, it is also popular as an edible medicinal fruit with names as “Buddha hand citron,” “Longevity orange,” and “Five finger orange” (Peng et al. 2009), and its name is usually misapplied as bergamot ( Citrus bergamia Risso) (Peng, Luo, et al. 2019; Peng, Yang, et al. 2019), nevertheless, their appearance is quite different. And they are two different kinds of fruits in Citrus genus (Donna et al. 2011). CF fruit has been utilized in some ingredients of Traditional Chinese Medicine (TCM) for the treatment of various chronic diseases including hypertension, respiratory tract infections, and type 2 diabetes mellitus (Peng et al. 2009). The bioactivities of CF are attributed to its bioactive compounds such as terpenes, flavonoids, coumarins and polysaccharides (Wu et al. 2013). The volatile compounds and essential oils of CF have drawn great attention of the researchers (Gao et al. 2020; Luo et al. 2019). Chemical composition and bioactivities of essential oils from CF have been studied in detail from all respects (Mahdi, Al‐Ansi, Ahmed, et al. 2019; Li et al. 2019; Xu et al. 2019). Besides, isolation, structural characterization and immunoregulatory activities of polysaccharides from CF have also been examined (Peng, Luo, et al. 2019; Peng, Yang, et al. 2019). However, the investigation of other chemical components such as flavonoids and coumarins which are also highly available in CF is limited (Chen et al. 2020). Recent comparative metabolomic and quantitative phytochemical studies have begun to broaden this characterization, expanding the chemical map of CF and linking specific flavonoids and coumarins to growth‐period‐dependent antioxidant capacity (El Sayed et al. 2024; Tan et al. 2025). Hence, there is a need to conduct comprehensive investigation for the chemical constituents of CF fruits. However, a comparative metabolomic investigation covering the major Chinese production regions of CF remains lacking. In addition, the potential of metabolomic profiling to distinguish CF from chayote ( Sechium edule ) has not been systematically evaluated. Although S. edule is taxonomically distant from C. medica , it is known in Chinese as “Foshougua,” whereas CF is commonly known as “Foshou.” This partially overlapping vernacular terminology, together with potential superficial confusion during commodity identification, makes S. edule a practically relevant comparator for a targeted authentication analysis. The purpose of such a comparison is not to treat the two species as biologically equivalent, but to evaluate whether their expected chemical differences can provide reproducible features for commodity discrimination.

In China, CF fruits are cultivated in different regions, and each of them develops its own regionally distinct types with different fruit morphologies. To date, there are four main production regions of CF in China. Cultivated in the provinces of Guangdong and Guangxi, CF is called “Guang” CF, and that grown in Sichuan province is named “Chuan” CF, and in the provinces of Yunnan, they are called “Yun” CF, respectively. In addition, “Jin” CF is named according to the main production city of Jinhua in Zhejiang province.

In the present study, CF samples from Zhejiang, Sichuan, Guangdong/Guangxi, and Yunnan were designated CF1, CF2, CF3, and CF4, respectively. Chayote ( S. edule , CFM) was selected a priori as the sole non‐Citrus comparator because it represents the specific commodity‐identification issue addressed in this study, rather than because it is taxonomically or biologically equivalent to CF. Untargeted UPLC‐Q‐Orbitrap MS/MS metabolomics, together with PCA, PLS‐DA, OPLS‐DA, hierarchical clustering, and pathway analysis, was used to characterize the metabolite profiles of these samples. The objectives were to (i) assess regional chemical variation among the four CF groups, (ii) identify candidate metabolites associated with regional quality discrimination, and (iii) evaluate whether metabolite fingerprints could support targeted authentication of CF against chayote. The conclusions concerning authentication are restricted to the sample groups examined in this study and require further external validation.

2. Materials and Methods

2.1. Plant Materials and Chemical Reagents

CF ( C. medica ) fruits were purchased from the local markets at different regions of China. Six biological replicates were analyzed for each sample group (CF1–CF4 and CFM). The fruits were chopped into small blocks, then they were frozen in liquid nitrogen immediately, and stored in −80°C for further analysis. The botanical identity of C. medica var. sarcodactylis and S. edule samples was confirmed by the corresponding author (R.F.) based on diagnostic morphological characteristics of the fruits; no voucher specimens were deposited because the samples were purchased as market commodities.

Acetonitrile and methanol with purity higher than 99.0% were purchased from Thermo Fisher Scientific Inc. (Shanghai, China). 2‐chlorophenylalanine was obtained from Aladdin Reagent Co. Ltd., China. Formic acid of LC–MS grade was obtained from TCI Shanghai. Ammonium formate was bought from Sigma‐Aldrich Inc. (St. Louis, MO, USA).

2.2. Extraction Procedure

Briefly, accurately weigh 100 mg (±1%) of tested sample, and add 1.2 mL 2‐chlorophenylalanine (4 ppm) methanol (precooled at −20°C), vortex for 30 s. Then 100 mg glass beads were added and the mixtures were ground at 25 Hz for 60 s in a tissue grinding machine (TissueLysis II). Subsequently, the samples were ultrasonicated at room temperature for 15 min. Afterwards, centrifugation was conducted at 25°C for 10 min at 1750 g, and the supernatant was filtered through a 0.22 μm membrane to obtain the prepared samples for LC–MS. The extract of 20 μL from each sample was mixed all together to make quality control (QC) samples.

2.3. UPLC‐Q‐Orbitrap MS/MS and Data Analysis

Chromatographic separation was performed on a Vanquish UHPLC system (Thermo Fisher Scientific) coupled to a Q Exactive HF‐X mass spectrometer (Thermo Fisher Scientific) operating in full MS/data‐dependent MS2 (ddMS2) mode. Separation was achieved on a Waters ACQUITY UPLC HSS T3 column (100 mm × 2.1 mm, 1.8 μm) maintained at 40°C. Mobile phase A consisted of 0.1% (v/v) formic acid in water containing 5 mmol/L ammonium formate, and mobile phase B was 0.1% formic acid in acetonitrile. The gradient elution programme was as follows: 0–2 min, 5% B; 2–12 min, 5%–90% B; 12–14 min, 90% B; 14–14.1 min, 90%–5% B; 14.1–17 min, 5% B (re‐equilibration). The flow rate was 0.3 mL/min and the injection volume was 2 μL. Mass spectrometry parameters were set as follows: spray voltage, +3.5/−2.8 kV (positive/negative mode); capillary temperature, 320°C; sheath gas flow rate, 45 arb; auxiliary gas, 10 arb; full MS scan range, m/z 100–1200; MS2 resolution, 17,500; normalized collision energy (NCE), 20, 40, and 60 eV (stepped). Peak identification, filtration, and alignment were conducted as described previously (Fan et al. 2020). Identification, filtration and alignment of peaks were conducted via XCMS (www.bioconductor.org) using R language (Smith et al. 2006). Data array of mass to charge ratio (m/z), retention time and peak intensity was subsequently obtained, and 13,947 precursor molecules in positive ion mode and 5435 precursors in negative ion mode were acquired for further analysis. Annotation of each metabolite was carried out according to their exact molecular weights and fragmentation pattern compared with the compounds in public spectral libraries, namely mzCloud (https://www.mzcloud.org), the Human Metabolome Database (HMDB), Metlin, MassBank and Lipid Maps, together with an in‐house spectral database built by Suzhou Bionovogene Co. Ltd. (Fan et al. 2020). All metabolite annotations reported in this study are based on MS2 spectral‐library matching without confirmation using authentic chemical standards. According to the Metabolomics Standards Initiative (MSI) guidelines (Sumner et al. 2007), these annotations correspond to confidence Level 2 (putatively annotated compounds).

2.4. Multivariate Statistical Analysis

The output data array from XCMS was processed in Microsoft Excel (Microsoft, Redmond, WA, USA) and all the data were mean‐centered and unit variance (UV)‐scaled for multivariate statistical analysis including principal component analysis (PCA), partial least squares‐discriminant analysis (PLS‐DA), and orthogonal partial least squares discriminant analysis (OPLS‐DA). All data were mean‐centered and UV‐scaled before multivariate statistical analysis. Sevenfold cross‐validation was used to assess the supervised models, and the PLS‐DA model was additionally evaluated using 100 random permutation tests as an internal diagnostic of model stability. Then, the normalized data was further processed by Simca‐P software version 13.0 (Umetrics AB, Umea, Sweden, www.umetrics.com/simca) and ropls package (Thévenot et al. 2015). Variables with a VIP > 1 and p < 0.05 were considered as significantly different.

2.5. Hierarchical Clustering and KEGG Enrichment Analysis

The differential metabolites were used to make the agglomerate hierarchical clustering analysis (HCA). The data arrays were processed by the program of pheatmap in the ropls package.

MetPA (MetaboAnalyst 5.0, www.metaboanalyst.ca) was applied to conduct the pathway analysis based on the KEGG annotation (https://www.kegg.jp/) through integrating pathway enrichment analysis and pathway topology analysis. The analysis result was visualized in a bubble plot. The higher the Impact and the lower the p value, the more significant the pathway. Raw pathway‐enrichment p values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) procedure implemented in MetaboAnalyst 5.0, and an FDR‐adjusted p < 0.05 was considered statistically significant.

3. Results and Discussion

3.1. Phytochemical Profiles

Before test of the samples, QC and quality assurance (QA) were carried out to make sure the reliability of methodology and data as shown in Figure S1. Theoretically, QC samples are all the same, but there will be systematic errors in the sample extraction, detection and analysis process, resulting in differences between QC samples. The smaller the difference, the higher the stability of the method, the better the data quality, which is reflected in PCA analysis as illustrated in Figure S1A indicating that the data are reliable (Dunn et al. 2011). In order to discover biomarkers, the relative standard deviation (RSD) of potential characteristic peaks in QC samples should not exceed 30% or the relevant characteristic peaks should be deleted (Dunn et al. 2011) which is the QA. In Figure S1B, the proportion of characteristic peaks with RSD < 30% is higher than 75% which further validate the reliability of the obtained data (Want et al. 2010).

Figure 1A,B show the typical base peak chromatograms of four regional CF groups and chayote (CFM) in positive and negative ion modes, respectively. After processing the raw data using ProteoWizard Data Analysis software (v3.0.8789) and XCMS, 13,947 and 5435 precursor features were detected in positive‐ and negative‐ion modes, respectively. A total of 380 metabolites, including both primary and secondary metabolites, were putatively annotated across the five sample groups (Table S1), as summarized in Figure 1C. Figure 1D shows the ten most represented chemical classes among the annotated metabolites. It is worth noting that flavonoids and phenols accounted for 3.68% and 3.42% of all annotated metabolites, respectively. These two types of compounds were proven to be bioactive with multiple functions in terms of protecting plants from stress and providing additional health benefits for human beings (Dixon et al. 2002).

FIGURE 1.

FIGURE 1

Open in a new tab

Base peak ion (BPI) chromatograms of four regional fingered citron groups (CF1–CF4) and chayote (CFM) in positive (A) and negative (B) modes, and classification of the metabolites detected into major functional classes (C) and top 10 chemical classes (D).

To further examine the flavonoids and phenols in CF fruits, all the secondary metabolites were selected and shown in Table 1. A total of 47 secondary metabolites including 23 flavonoids, 8 terpenoids, 8 phenols, 3 coumarins, 3 alkaloids, and 2 steroids were putatively annotated by UPLC‐Q‐Orbitrap MS/MS. In order to compare the contents of secondary metabolites in different regional groups, a rank for the relative amount of individual compounds was performed and the results were presented in Table 1. CF2 from Sichuan and CF1 from Zhejiang showed the two highest cumulative relative‐abundance rankings for the secondary metabolites listed in Table 1. In contrast, CFM showed the lowest overall relative abundance of these secondary metabolites, whereas CF3 and CF4 displayed intermediate profiles.

TABLE 1.

Main secondary metabolites and their relative‐abundance rankings in four regional fingered citron groups (CF1–CF4) and chayote (CFM).

No. Name Classification CF1 CF2 CF3 CF4 CFM
1 Hesperetin Flavanone 3 5 2 4 1
2 Neohesperidin Flavanone 7‐O‐neohesperidose 3 5 2 4 1
3 Chrysin* Flavone 5 4 1 2 3
4 Luteolin* Flavone 5 3 4 2 1
5 Rhoifolin Flavone 7‐O‐neohesperidose 4 5 3 2 1
6 Scolymoside Flavone 7‐O‐neohesperidose 5 3 1 4 2
7 Isovitexin* Flavone 6‐C‐glucoside 2 5 1 3 4
8 Apiin Flavone diglycoside 5 4 3 2 1
9 Quercetin* Flavonol 5 4 3 1 2
10 Kaempferol Flavonol 4 1 3 2 5
11 Isoquercitrin* Flavonol 3‐O‐glucoside 5 4 2 1 3
12 Rutin Flavonol 3‐O‐rutinoside 5 4 3 2 1
13 Kaempferide O‐methylated flavonol 3 5 2 4 1
14 3‐O‐methylquercetin* O‐methylated flavonol 5 4 2 3 1
15 Pelargonin* Anthocyanin 5 4 3 2 1
16 Leucopelargonidin Anthocyanin 5 4 3 2 1
17 Peonidin‐3‐glucoside* Anthocyanin 3 5 2 4 1
18 Malvidin 3‐glucoside Anthocyanin 4 2 5 1 3
19 Cyanidin 3‐glucoside Anthocyanin 3 4 1 2 5
20 Cyanidin 3‐O‐rutinoside 5‐O‐beta‐d‐glucoside* Anthocyanin 5 3 1 2 4
21 Cyanidin 3‐O‐(2‐O‐beta‐D‐glucuronosyl)‐beta‐D‐glucoside Anthocyanin 4 5 3 1 2
22 Delphinidin 3‐O‐beta‐D‐glucoside 5‐O‐(6‐coumaroyl‐beta‐d‐glucoside)* Anthocyanin 5 3 1 2 4
23 (−)‐Epigallocatechin* Flavan‐3‐ol 2 3 5 4 1
24 Pulegone Monoterpene 2 5 4 3 1
25 d‐Limonene Monoterpene 5 3 2 4 1
26 (+)‐α‐Pinene* Monoterpene 3 5 4 2 1
27 (1S,4R)‐1‐Hydroxy‐2‐oxolimonene Monoterpene derivative 3 5 2 4 1
28 Limonene‐1,2‐diol* Monoterpene alcohol 2 4 3 5 1
29 2‐Trans,6‐trans‐farnesal* Sesquiterpene alcohol 3 5 4 2 1
30 (1R,4R)‐Dihydrocarvone Ketone 4 2 1 3 5
31 Eucalyptol Monoterpene ether 2 4 5 1 3
32 Sinapyl alcohol Phenol 5 4 3 2 1
33 Perillyl alcohol* Phenol 3 5 2 4 1
34 Coniferyl alcohol* Phenol 5 3 4 2 1
35 Chavicol Phenylpropene 1 4 3 2 5
36 Eugenol* Allylbenzene 4 5 3 2 1
37 Isochavicol Phenol 1 4 3 2 5
38 Methyleugenol Phenylpropene 4 2 5 3 1
39 Estragole Phenylpropene 5 3 4 2 1
40 Coumarin Coumarin 3 2 4 5 1
41 Scopoletin Coumarin 4 5 3 2 1
42 Umbelliferone* Coumarin 5 4 3 2 1
43 N‐methylserotonin* Alkaloid 5 4 3 2 1
44 N‐acetylserotonin Alkaloid 2 5 1 3 4
45 Deoxyloganin Alkaloid 5 3 4 2 1
46 Campestanol Steroid 2 3 1 4 5
47 β‐Sitosterol Steroid 2 5 4 1 3
Total 175 183 131 120 96
Open in a new tab
*

Nineteen secondary metabolites classified as differential metabolites using VIP > 1 and p < 0.05.

In various different types of secondary metabolites, flavonoids cover almost half of the total secondary metabolites which indicated that CF fruits are abundant of flavonoids including flavanones, flavones, flavonols and anthocyanins. Some of the flavonoids were also reported to be contained in other Citrus family such as hesperetin, rhoifolin, neohesperidin, luteolin, quercetin, rutin (Soares et al. 2015; Zhang et al. 2012; Olas 2021; Johnson et al. 2011; Fan et al. 2019). These citrus flavonoids exert good physiological activities such as antiproliferative, antitumor (Ferreira de Oliveira et al. 2020), antidiabetic (Rao et al. 2011), neuroprotective (Brinza et al. 2020), prebiotic (Lu et al. 2020) activities. Some of the secondary metabolites such as 3‐O‐methylquercetin, pelargonin, leucopelargonidin, peonidin‐3‐glucoside, malvidin 3‐glucoside (1S,4R)‐1‐hydroxy‐2‐oxolimonene, sinapyl alcohol, isochavicol, N‐acetylserotonin, deoxyloganin, campestanol, and β‐sitosterol were tentatively annotated in CF for the first time, to the best of our knowledge. Thus, this untargeted metabolomic analysis expands the available chemical information on CF and provides a basis for its further nutritional and medicinal evaluation. It should be noted that all metabolite annotations were based on MS2 spectral‐library matching without confirmation using authentic chemical standards, corresponding to MSI confidence Level 2; therefore, the structural assignments of these tentatively annotated compounds should be interpreted with appropriate caution.

3.2. Multivariate Analysis of Annotated Metabolites

PCA was first conducted to evaluate the overall variation in the annotated metabolite profiles without using prior group information. As shown in Figure 2A, CF2, CF3, and CF4 clustered relatively closely, whereas CF1 was more clearly separated from these three regional groups. CFM was distinctly separated from all four CF groups. Given that C. medica and S. edule belong to taxonomically distant families and orders, this interspecific separation was expected and should not be interpreted as an unexpected biological finding. The observed distribution may reflect the combined effects of taxonomic identity, regional origin, genotype, and environmental conditions, which could not be independently disentangled in the present sampling design.

FIGURE 2.

FIGURE 2

Open in a new tab

(A) Principal component analysis (PCA), (B) permutations plot of the PLS‐DA model, (C) partial least squares‐discriminant analysis (PLS‐DA), and (D) orthogonal projections to latent structures discriminant analysis (OPLS‐DA) of four regional fingered citron groups (CF1–CF4) and chayote (CFM) for the CF1 versus CF2 versus CF3 versus CF4 versus CFM in positive ion mode.

PLS‐DA and OPLS‐DA were subsequently used to further characterize the discrimination among the sample groups. The PLS‐DA score plot (Figure 2C) showed a grouping pattern broadly consistent with that observed in the unsupervised PCA. The 100‐permutation test was used as an internal diagnostic of PLS‐DA model stability (Figure 2B) (Wheelock and Wheelock 2013). The OPLS‐DA models also showed clear separation among the analyzed groups (Figure 2D; Table 2), with R 2 Y values ranging from 0.996 to 1.000 and Q 2 values ranging from 0.984 to 0.994. These exceptionally high Q 2 values indicate strong cross‐validated separation within the present dataset but should be interpreted cautiously. In particular, the large taxonomic and metabolic distance between CFM and the CF groups may contribute substantially to the observed separation, while the limited number of biological replicates may result in optimistic estimates of predictive performance. Although the permutation analysis and the generally consistent grouping observed in PCA provide internal support for the model stability, these procedures cannot completely exclude model overfitting or replace validation using an independent external sample set. Therefore, the high Q 2 values should be interpreted as evidence of strong separation within the present dataset rather than definitive evidence of external predictive accuracy.

TABLE 2.

Values of the statistic parameters obtained for different OPLS‐DA models based on LC‐MS data (in positive mode). a

Model classes OPLS‐DA Number of DM
Pre R 2 X (cum, %) R 2 Y (cum, %) Q 2 (cum, %)
CF2 vs. CF3 2 55.7 100 98.5 188
CF2 vs. CF4 2 59.0 100 99.0 201
CF2 vs. CFM 2 65.3 100 99.4 231
CF2 vs. CF1 2 54.7 100 98.5 185
CF3 vs. CF4 2 54.9 100 98.7 191
CF3 vs. CFM 2 62.5 100 99.3 225
CF3 vs. CF1 2 58.5 100 99.0 199
CF4 vs. CFM 2 63.8 100 99.3 244
CF4 vs. CF1 2 60.7 100 99.1 200
CFM vs. CF1 2 67.1 100 99.4 248
CF1 vs. CF2 vs. CF3 vs. CF4 vs. CFM 2 31.7 99.6 98.4 139
Open in a new tab
a

Pre is the number of principal components; R 2 X (cum), cumulative fraction of X variance explained by the model; R 2 Y (cum), cumulative fraction of Y variance explained by the model; Q 2 (cum), cumulative cross‐validated predictive ability; DM, differential metabolite selected using VIP > 1 and p < 0.05. Detailed information on the DMs is provided in Table S2.

A total of 65 differential metabolites were identified among the five sample groups, including 19 secondary metabolites comprising 11 flavonoids, three terpenoids, three phenols, one coumarin, and one alkaloid. The numbers of differential metabolites detected in the pairwise comparisons are summarized in Table S2, with the largest number observed in the CFM versus CF1 comparison. These compounds are regarded as candidate discriminatory metabolites rather than validated biomarkers because their predictive performance has not yet been evaluated using an independent sample set.

Importantly, C. medica belongs to Rutaceae within Sapindales, whereas S. edule belongs to Cucurbitaceae within Cucurbitales. Their marked metabolic divergence is therefore expected and is consistent with the evolutionary and phylogenetic structuring of plant secondary metabolism (Wink 2003; Vieira et al. 2019). Accordingly, the comparison between the two species was not intended as a biologically equivalent or phylogenetically matched contrast. Instead, S. edule was included as a targeted authentication comparator because the partially overlapping Chinese vernacular terms “Foshou” and “Foshougua” may contribute to commodity misidentification. Within this authentication‐oriented framework, the interspecific differential metabolites provide candidate chemical features for distinguishing the two commodities, whereas the differences among CF1–CF4 may support regional quality discrimination. This interpretation is consistent with current LC‐MS‐based food‐authentication strategies (Zhong et al. 2022) and metabolomic approaches used to discriminate citrus materials (Song et al. 2025). Nevertheless, the authentication scope of the present study is restricted to the sample groups examined here and requires further validation using independent external samples.

3.3. HCA

A total of 65 differential metabolites were used for HCA. As shown in Figure 3, their relative abundance patterns differed among the four regional CF groups and CFM. Thirty‐four of these metabolites showed relatively higher abundance in CFM, whereas the remaining metabolites displayed variable distributions among CF1–CF4. The HCA grouping was generally consistent with the PCA results. Within the present sample set, these 65 differential metabolites represent candidate chemical features for distinguishing CF from the targeted chayote comparator, while the metabolites differing among CF1–CF4 may contribute to regional quality discrimination. However, they should not yet be regarded as validated authentication biomarkers because their sensitivity and specificity have not been evaluated using independent samples.

FIGURE 3.

FIGURE 3

Open in a new tab

Heatmap of hierarchical clustering analysis of differential metabolites among four regional fingered citron groups (CF1–CF4) and chayote (CFM). The abscissa indicates different groups labeled with different colors for the main groups and numerically marked for the subgroups. The ordinate indicates the differential metabolites selected as candidate discriminatory metabolites in four regional fingered citron groups (CF1–CF4) and chayote (CFM). The bar at the right of the heat map represents row‐scaled relative metabolite abundances.

Since secondary metabolites are important bioactive components of fruits, we further used 19 secondary differential metabolites to conduct the HCA and present the result in Figure S2. Several secondary metabolites, including eight flavonoids, one phenol, one alkaloid, and one coumarin, showed relatively high abundance in CF1. This result suggests that the Zhejiang sample group possesses a distinctive secondary‐metabolite profile. However, metabolite abundance alone is insufficient to establish overall commercial or medicinal quality, and targeted quantification, sensory evaluation, bioactivity assays, and broader multi‐year sampling are required before drawing conclusions regarding regional quality superiority.

3.4. KEGG Annotation and Metabolic Pathway Analysis

Plant metabolites perform diverse physiological, ecological, and adaptive functions (Luo 2015). Unraveling the metabolic pathways and networks will provide information for the genetic basis of plant metabolism (Carreno‐Quintero et al. 2013). As noted above, both the taxonomic origin (species) and the growing environment influence the accumulation of plant secondary metabolites. In the present study, we compared the metabolite profiles of CF from the different regions and of chayote. The differential metabolites from the comparison were annotated by the Kyoto Encyclopedia of Genes and Genomes (KEGG), a database that integrates genomic, chemical, and systemic functional information (Kanehisa and Goto 2000). The pathway analysis was conducted by MetaboAnalyst (https://www.metaboanalyst.ca/, version 5.0) through integrating pathway enrichment analysis and pathway topology analysis. Figure 4A presents a bubble plot visualizing the pathway analysis of differential metabolites in four regional CF groups (CF1–CF4) and chayote (CFM). The enrichment analysis indicates that pathway of “Flavone and flavonol biosynthesis” has the highest impact (> 0.6) and the p value of “Phenylalanine metabolism” pathway is the least (< 0.01). Therefore, although neither pathway remained significant after FDR correction, these two pathways showed the strongest nominal enrichment for the differential metabolites from the comparison of five samples. Figure 4B summarizes all the tested pathways as a horizontal bar chart, in which the bar length represents the pathway impact value obtained from the pathway topology analysis and the bar color represents the corresponding raw (nominal) p value; the pathways are ordered from top to bottom by increasing raw p value (i.e., decreasing nominal significance).

FIGURE 4.

FIGURE 4

Open in a new tab

Pathway analysis of differential metabolites in four regional fingered citron groups (CF1–CF4) and chayote (CFM) illustrated as a bubble diagram (A) and histogram (B). In (A), the y‐axis and bubble color denote −log10 (raw, nominal p), bubble size denotes the number of hits, and pathway impact is shown on the x‐axis; the dashed line marks nominal p = 0.05. In (B), bar length denotes the pathway impact value obtained from the pathway topology analysis, and bar color denotes the raw (nominal) p value of the pathway enrichment test, with paler colors indicating lower p values and darker colors indicating higher p values (color scale at right); pathways are arranged from top to bottom in order of increasing raw p value.

4. Conclusions

In this study, untargeted UPLC‐Q‐Orbitrap MS/MS metabolomics was used to characterize the phytochemical profiles of four regional CF groups and chayote, which was included as a targeted non‐Citrus authentication comparator. A total of 380 metabolites were putatively annotated, including 47 secondary metabolites. Twelve secondary metabolites were tentatively annotated in CF for the first time, to the best of our knowledge. PCA, PLS‐DA, OPLS‐DA, and hierarchical clustering revealed clear differences among the five sample groups and identified 65 differential metabolites, including 19 secondary metabolites. The separation between C. medica and S. edule was expected because of their substantial taxonomic and metabolic distance. Therefore, this interspecific comparison should be interpreted as a targeted chemical‐authentication analysis rather than a biologically equivalent comparison. The differences among CF1–CF4 additionally provide preliminary information for regional chemical‐quality discrimination. Phenylalanine metabolism and flavone and flavonol biosynthesis showed the strongest enrichment at the nominal p value level, although neither pathway remained significant after FDR correction. Consequently, these pathway results should be considered exploratory. Overall, the study provides a metabolite reference for regional CF samples and identifies candidate discriminatory features with potential utility in product authentication and regional quality evaluation. However, the proposed features are not yet validated authentication biomarkers. The study was limited by the inclusion of only one non‐Citrus comparator, a limited number of biological replicates, single‐season sampling, the reliance on database‐level spectral matching without authentic‐standard verification (MSI Level 2), and the absence of independent external validation and direct bioactivity assessment. Future studies should include additional Citrus taxa and potentially confused commodities, multiple harvest years and production sites, targeted quantitative validation, and independent sample sets to determine the specificity, reproducibility, and practical applicability of the candidate markers.

Author Contributions

Ruiyi Fan: investigation, validation, formal analysis, supervision, funding acquisition, writing – review and editing. Bin Li: conceptualization, methodology, software, data curation, writing – original draft.

Funding

This work was financially supported by the PhD Startup Foundation of Jingchu University of Technology (YYZ202512), the Key Project of the Science and Technology Research Program of the Department of Education of Hubei Province (D20244303), and the Jingmen City Innovation and Development Joint Fund of the Hubei Provincial Natural Science Foundation (JCZRLH202601519).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Score scatter plots for principal component analysis (PCA) model with all the samples represented by green dots and the quality control (QC) samples represented by red dots in positive mode (A) and negative mode (B). Relative standard deviation (RSD) tests of the detected peaks in positive mode (C) and negative mode (D).

Figure S2: Heatmap of hierarchical clustering analysis of 19 secondary differential metabolites. The abscissa indicates different groups labeled with different color for the main groups and numerically marked for the subgroups. The ordinate indicates the differential metabolites selected as candidate discriminatory metabolites in four regional fingered citron groups (CF1–CF4) and chayote (CFM). The bar at the right of the heat map represents relative expression values.

FSN3-14-e72115-s003.docx (210.8KB, docx)

Table S1: All compounds identified in this study.

FSN3-14-e72115-s001.xlsx (59.6KB, xlsx)

Table S2: The information of the differential metabolites compared in various groups (information of each group was provided in individual sheets).

FSN3-14-e72115-s002.xlsx (284.5KB, xlsx)

Acknowledgments

The authors gratefully acknowledge the support from the Plant Biological Breeding Team. During the preparation of this manuscript, the authors used AI‐assisted tools solely to improve the language and readability of the text. After using these tools, the authors reviewed and edited the content as needed and took full responsibility for the content of the publication. No generative AI was used to generate, analyze, or interpret the experimental data or the scientific results.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

  1. Brinza, I. , Abd‐Alkhalek A. M., El‐Raey M. A., Boiangiu R. S., Eldahshan O. A., and Hritcu L.. 2020. “Ameliorative Effects of Rhoifolin in Scopolamine‐Induced Amnesic Zebrafish ( Danio rerio ) Model.” Antioxidants 9: 580. 10.3390/antiox9070580. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Carreno‐Quintero, N. , Bouwmeester H. J., and Keurentjes J. J.. 2013. “Genetic Analysis of Metabolome‐Phenotype Interactions: From Model to Crop Species.” Trends in Genetics 29: 41–50. 10.1016/j.tig.2012.09.006. [DOI] [PubMed] [Google Scholar]
  3. Chen, Q. Y. , Wang D., Tan C., Hu Y., Sundararajan B., and Zhou Z.. 2020. “Profiling of Flavonoid and Antioxidant Activity of Fruit Tissues From 27 Chinese Local Citrus Cultivars.” Plants 9, no. 2: 196. 10.3390/plants9020196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Dixon, R. A. , Achnine L., Kota P., Liu C. J., and Wang L.. 2002. “The Phenylpropanoid Pathway and Plant Defence—A Genomics Perspective.” Molecular Plant Pathology 3, no. 5: 371–390. 10.1046/j.1364-3703.2002.00131.x. [DOI] [PubMed] [Google Scholar]
  5. Donna, L. D. , Gallucci G., Malaj N., Romano E., Tagarelli A., and Sindona G.. 2011. “Recycling of Industrial Essential Oil Waste: Brutieridin and Melitidin, Two Anticholesterolaemic Active Principles From Bergamot Albedo.” Food Chemistry 125, no. 2: 438–441. 10.1016/j.foodchem.2010.09.025. [DOI] [Google Scholar]
  6. Dunn, W. B. , Broadhurst D., Begley P., et al. 2011. “Procedures for Large‐Scale Metabolic Profiling of Serum and Plasma Using Gas Chromatography and Liquid Chromatography Coupled to Mass Spectrometry.” Nature Protocols 6: 1060–1083. 10.1038/nprot.2011.335. [DOI] [PubMed] [Google Scholar]
  7. El Sayed, A. M. , El‐Abd E. A. W., Afifi A. H., et al. 2024. “Comparative Metabolomics Analysis of Citrus medica var. sarcodactylis Swingle and Limonia acidissima Linn. Fruits and Leaves Cultivated in Egypt in Context to Their Antiviral Effects.” Heliyon 10: e32335. 10.1016/j.heliyon.2024.e32335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Fan, R. , Peng C., Zhang X., et al. 2020. “A Comparative UPLC‐Q‐Orbitrap‐MS Untargeted Metabolomics Investigation of Different Parts of Clausena lansium (Lour.) Skeels.” Food Science & Nutrition 8: 5811–5822. 10.1002/fsn3.1841. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Fan, R. , Zhu C., Qiu D., and Zeng J.. 2019. “Comparison of the Bioactive Chemical Components and Antioxidant Activities in Three Tissues of Six Varieties of Citrus grandis ‘Tomentosa’ Fruits.” International Journal of Food Properties 22: 1848–1862. 10.1080/10942912.2019.1683027. [DOI] [Google Scholar]
  10. Ferreira de Oliveira, J. M. P. , Santos C., and Fernandes E.. 2020. “Therapeutic Potential of Hesperidin and Its Aglycone Hesperetin: Cell Cycle Regulation and Apoptosis Induction in Cancer Models.” Phytomedicine 73: 152887. 10.1016/j.phymed.2019.152887. [DOI] [PubMed] [Google Scholar]
  11. Gao, Z. , Zhong W., Chen K., Tang P., and Guo J.. 2020. “Chemical Composition and Anti‐Biofilm Activity of Essential Oil From Citrus medica L. var. sarcodactylis Swingle Against Listeria monocytogenes .” Industrial Crops and Products 144: 112036. 10.1016/j.indcrop.2019.112036. [DOI] [Google Scholar]
  12. Johnson, J. L. , Rupasinghe S. G., Stefani F., Schuler M. A., and Gonzalez de Mejia E.. 2011. “ Citrus Flavonoids Luteolin, Apigenin, and Quercetin Inhibit Glycogen Synthase Kinase‐3beta Enzymatic Activity by Lowering the Interaction Energy Within the Binding Cavity.” Journal of Medicinal Food 14: 325–333. 10.1089/jmf.2010.0310. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Kanehisa, M. , and Goto S.. 2000. “KEGG: Kyoto Encyclopedia of Genes and Genomes.” Nucleic Acids Research 28, no. 1: 27–30. 10.1093/nar/28.1.27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Li, Z.‐h. , Cai M., Yang K., and Sun P.‐l.. 2019. “Kinetic Study of d‐Limonene Release From Finger Citron Essential Oil Loaded Nanoemulsions During Simulated Digestion In Vitro.” Journal of Functional Foods 58: 67–73. 10.1016/j.jff.2019.04.056. [DOI] [Google Scholar]
  15. Lu, J. F. , Zhu M. Q., Zhang H., et al. 2020. “Neohesperidin Attenuates Obesity by Altering the Composition of the Gut Microbiota in High‐Fat Diet‐Fed Mice.” FASEB Journal: Official Publication of the Federation of American Societies for Experimental Biology 34: 12053–12071. 10.1096/fj.201903102RR. [DOI] [PubMed] [Google Scholar]
  16. Luo, C. , Li D., Wang Y., Guo S., Zhang D., and Du S.. 2019. “Chemical Composition and Insecticide Efficacy of Essential Oils From Citrus medica L. var. sarcodactylis Swingle Against Tribolium castaneum Herbst in Stored Medicinal Materials.” Journal of Essential Oil Bearing Plants 22: 1182–1194. 10.1080/0972060x.2019.1685914. [DOI] [Google Scholar]
  17. Luo, J. 2015. “Metabolite‐Based Genome‐Wide Association Studies in Plants.” Current Opinion in Plant Biology 24: 31–38. 10.1016/j.pbi.2015.01.006. [DOI] [PubMed] [Google Scholar]
  18. Mahdi, A. A. , Al‐Ansi W., Ahmed M. I., et al. 2019. “Microwave Assisted Extraction of the Bioactive Compounds From Peel/Pulp of Citrus medica L. var. sarcodactylis Swingle Along With Its Nutritional Profiling.” Journal of Food Measurement and Characterization 14: 283–292. 10.1007/s11694-019-00290-6. [DOI] [Google Scholar]
  19. Mahdi, A. A. , Al‐Ansi W., Al‐Maqtari Q. A., Ahmed A., and Wang H.. 2019. “GC‐MS Analysis of Volatile Compounds the Peel and Pulp of Citrus medica L. var. sarcodactylis Swingle (Foshou Fruit).” International Journal of Research in Agricultural Sciences 6: 134–142. [Google Scholar]
  20. Olas, B. 2021. “A Review of In Vitro Studies of the Anti‐Platelet Potential of Citrus Fruit Flavonoids.” Food and Chemical Toxicology 150: 112090. 10.1016/j.fct.2021.112090. [DOI] [PubMed] [Google Scholar]
  21. Peng, B. , Luo Y., Hu X., et al. 2019. “Isolation, Structural Characterization, and Immunostimulatory Activity of a New Water‐Soluble Polysaccharide and Its Sulfated Derivative From Citrus medica L. var. sarcodactylis .” International Journal of Biological Macromolecules 123: 500–511. 10.1016/j.ijbiomac.2018.11.113. [DOI] [PubMed] [Google Scholar]
  22. Peng, B. , Yang J., Huang W., et al. 2019. “Structural Characterization and Immunoregulatory Activity of a Novel Heteropolysaccharide From Bergamot ( Citrus medica L. var. sarcodactylis) by Alkali Extraction.” Industrial Crops and Products 140: 111617. 10.1016/j.indcrop.2019.111617. [DOI] [Google Scholar]
  23. Peng, C. H. , Ker Y. B., Weng C. F., Peng C. C., and Peng R. Y.. 2009. “Insulin Secretagogue Bioactivity of Finger Citron Fruit ( Citrus medica l. var. sarcodactylis Hort, Rutaceae).” Journal of Agricultural and Food Chemistry 57, no. 19: 8812–8819. 10.1021/jf902143x. [DOI] [PubMed] [Google Scholar]
  24. Rao, Y. K. , Lee M. J., Chen K., Lee Y. C., Wu W. S., and Tzeng Y. M.. 2011. “Insulin‐Mimetic Action of Rhoifolin and Cosmosiin Isolated From Citrus grandis (L.) Osbeck Leaves: Enhanced Adiponectin Secretion and Insulin Receptor Phosphorylation in 3T3‐L1 Cells.” Evidence‐Based Complementary and Alternative Medicine 2011: 624375. 10.1093/ecam/nep204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Smith, C. A. , Want E. J., O'Maille G., Abagyan R., and Siuzdak G.. 2006. “XCMS: Processing Mass Spectrometry Data for Metabolite Profiling Using Nonlinear Peak Alignment, Matching, and Identification.” Analytical Chemistry 78: 779–787. 10.1021/ac051437y. [DOI] [PubMed] [Google Scholar]
  26. Soares, M. S. , da Silva D. F., Forim M. R., et al. 2015. “Quantification and Localization of Hesperidin and Rutin in Citrus sinensis Grafted on C. limonia After Xylella fastidiosa Infection by HPLC‐UV and MALDI Imaging Mass Spectrometry.” Phytochemistry 115: 161–170. 10.1016/j.phytochem.2015.02.011. [DOI] [PubMed] [Google Scholar]
  27. Song, Z. , Xu J., Tian J., et al. 2025. “Differentiating Tangerine Peels From Other Citrus reticulata Through GC‐MS, UPLC‐Q‐Exactive Orbitrap‐MS, and HPLC‐PDA.” ACS Omega 10: 1688–1704. 10.1021/acsomega.4c09701. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Sottile, F. , Signore M. B. D., and Barone E.. 2019. “Ornacitrus: Citrus Plants (Citrus spp.) as Ornamentals.” Folia Horticulturae 31: 239–251. 10.2478/fhort-2019-0018. [DOI] [Google Scholar]
  29. Sumner, L. W. , Amberg A., Barrett D., et al. 2007. “Proposed Minimum Reporting Standards for Chemical Analysis: Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI).” Metabolomics 3: 211–221. 10.1007/s11306-007-0082-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Tan, T. , Xu M., Hong X., et al. 2025. “Quantitative Analysis of Flavonoids and Coumarins From Fingered Citron in Different Growth Periods and Their Regulatory Effects on Oxidative Stress.” Foods 14: 180. 10.3390/foods14020180. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Thévenot, E. A. , Roux A., Xu Y., Ezan E., and Junot C.. 2015. “Analysis of the Human Adult Urinary Metabolome Variations With Age, Body Mass Index, and Gender by Implementing a Comprehensive Workflow for Univariate and OPLS Statistical Analyses.” Journal of Proteome Research 14: 3322–3335. 10.1021/acs.jproteome.5b00354. [DOI] [PubMed] [Google Scholar]
  32. Vieira, E. F. , Pinho O., Ferreira I. M. P. L. V. O., and Delerue‐Matos C.. 2019. “Chayote ( Sechium edule ): A Review of Nutritional Composition, Bioactivities and Potential Applications.” Food Chemistry 275: 557–568. 10.1016/j.foodchem.2018.09.146. [DOI] [PubMed] [Google Scholar]
  33. Want, E. J. , Wilson I. D., Gika H., et al. 2010. “Global Metabolic Profiling Procedures for Urine Using UPLC–MS.” Nature Protocols 5: 1005–1018. 10.1038/nprot.2010.50. [DOI] [PubMed] [Google Scholar]
  34. Wheelock, A. M. , and Wheelock C. E.. 2013. “Trials and Tribulations of 'omics Data Analysis: Assessing Quality of SIMCA‐Based Multivariate Models Using Examples From Pulmonary Medicine.” Molecular BioSystems 9: 2589–2596. 10.1039/c3mb70194h. [DOI] [PubMed] [Google Scholar]
  35. Wink, M. 2003. “Evolution of Secondary Metabolites From an Ecological and Molecular Phylogenetic Perspective.” Phytochemistry 64: 3–19. 10.1016/S0031-9422(03)00300-5. [DOI] [PubMed] [Google Scholar]
  36. Wu, Z. , Li H., Yang Y., Zhan Y., and Tu D.. 2013. “Variation in the Components and Antioxidant Activity of Citrus medica L. var. sarcodactylis Essential Oils at Different Stages of Maturity.” Industrial Crops and Products 46: 311–316. 10.1016/j.indcrop.2013.02.015. [DOI] [Google Scholar]
  37. Xu, Y. , Zhu C., Xu C., et al. 2019. “Integration of Metabolite Profiling and Transcriptome Analysis Reveals Genes Related to Volatile Terpenoid Metabolism in Finger Citron ( C. medica var. sarcodactylis).” Molecules 24: 1–21. 10.3390/molecules24142564. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Zhang, J. , Sun C., Yan Y., et al. 2012. “Purification of Naringin and Neohesperidin From Huyou (Citrus changshanensis) Fruit and Their Effects on Glucose Consumption in Human HepG2 Cells.” Food Chemistry 135: 1471–1478. 10.1016/j.foodchem.2012.06.004. [DOI] [PubMed] [Google Scholar]
  39. Zhong, P. , Wei X., Li X., et al. 2022. “Untargeted Metabolomics by Liquid Chromatography‐Mass Spectrometry for Food Authentication: A Review.” Comprehensive Reviews in Food Science and Food Safety 21: 2455–2488. 10.1111/1541-4337.12938. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S1: Score scatter plots for principal component analysis (PCA) model with all the samples represented by green dots and the quality control (QC) samples represented by red dots in positive mode (A) and negative mode (B). Relative standard deviation (RSD) tests of the detected peaks in positive mode (C) and negative mode (D).

Figure S2: Heatmap of hierarchical clustering analysis of 19 secondary differential metabolites. The abscissa indicates different groups labeled with different color for the main groups and numerically marked for the subgroups. The ordinate indicates the differential metabolites selected as candidate discriminatory metabolites in four regional fingered citron groups (CF1–CF4) and chayote (CFM). The bar at the right of the heat map represents relative expression values.

FSN3-14-e72115-s003.docx (210.8KB, docx)

Table S1: All compounds identified in this study.

FSN3-14-e72115-s001.xlsx (59.6KB, xlsx)

Table S2: The information of the differential metabolites compared in various groups (information of each group was provided in individual sheets).

FSN3-14-e72115-s002.xlsx (284.5KB, xlsx)

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Articles from Food Science & Nutrition are provided here courtesy of Wiley

RESOURCES