Chemical Compositions and Antioxidant Activities of Indonesian Citrus Essential Oils and Their Elucidation Using Principal Component Analysis

Citrus essential oils (EOs) have various bioactivities like antioxidants, with many applications. Antioxidant activities depend on the chemical compositions of the EOs, which are affected by climate, soil, and geographical region. Thus, investigations on chemical compositions and antioxidant activities of Citrus EOs in different countries are valuable. In this study, we distilled EOs from peels of Indonesian-grown Citrus, including C. nobilis, C. limon, C. aurantifolia, C. amblycarpa, and Citrus spp. Chemical compositions of EOs were analyzed using Gas Chromatography-Mass Spectrometer (GC-MS), whereas the antioxidant activities were determined by employing 2,2-diphenyl-2-picrylhydrazyl (DPPH) method. Furthermore, principal component analysis (PCA) was applied to elucidate the main contributing compounds for antioxidant activity. The results show that all EOs possess unique chemical characteristics, with limonene as the majority constituent. For antioxidant activities, C. limon and C. amblycarpa EOs are the two strongest, IC50 values below 7.00 μL/mL. PCA approach suggests that -terpinene mainly contributes to the high antioxidant activities of C. limon and C. amblycarpa. Moreover, o-cymene, thymol, p-cymene, and αpharnesene may also be responsible for the antioxidant activity of C. limon EO. These results are valuable information for the applications of Citrus EOs as antioxidant sources.

Citrus EOs have diverse bioactivities, such as antioxidant [6,7]. Antioxidant protects cellular structure and function by scavenging free radicals, inhibiting lipid peroxidation, and preventing other oxidative degradations [1]. Therefore, compounds with antioxidant bioactivities have been widely used as additives in food products [9] and antiaging in cosmetics [3].
Chemical compositions of Citrus EOs in terms of qualitative and quantitative are affected by climate, soil, and geographical region [13]. Meanwhile, chemical compositions of Citrus EOs determine their antioxidant activities [14]. Thus, investigating chemical compositions and antioxidant activities of Citrus EOs from other countries are important research topics. In this present work, we distilled Citrus EOs from peels of C. nobilis, C. limon, C. aurantifolia, C. amblycarpa, and Citrus spp. grown in Indonesia. Chemical compositions of Citrus EOs were analyzed using Gas Chromatography-Mass Spectrometer (GC-MS). Antioxidant activities of all EOs were also determined using 2,2-diphenyl-2-picrylhydrazyl (DPPH) method. We applied principal component analysis (PCA) to elucidate the responsible secondary metabolites for the antioxidant activity. PCA is a powerful tool for exploratory data analysis and has vital roles in many fields, such as drug design [15], chemotaxonomy [16], and metabolite profiling [17]. This study is essential in providing valuable information about the antioxidant activities of various Citrus EOs and the responsible secondary metabolites. This study will help the application of Citrus EOs in many fields, for example, health, food, functional fabric, and cosmetics.

Yields of Citrus Essential Oils
In this present work, we have distilled EOs from five Citrus species. The obtained Citrus EOs are clear liquids and have pale-yellow colors with fresh aromas. The resulting EO yields vary among species, ranging from 0.34-3.70%. The highest EO yield was acquired from the peel of C. nobilis (3.70%), followed by C. limon (0.85%), C. aurantifolia (0.78%), C. amblycarpa (0.65%), and Citrus spp. (0.34%). Our results are slightly higher than those reported by others [5,18], where yields of EOs are usually between 0.2 and 2.3%. The EO yields depend on plant genetics, climate, soil, geographical location, and storage period [19]. Therefore, the amount of EO obtained may vary even for the same species. Table 2 shows that the five Citrus EOs have different chemical compositions in terms of qualitative and quantitative (see Figure S1 for the chemical structures). The Citrus EOs contain monoterpene, monoterpenoid, sesquiterpene, and sesquiterpenoid types of secondary metabolites. These secondary metabolite classes sparsely distribute among the five Citrus EOs. Monoterpenes are the dominating secondary metabolites in Citrus EOs, where limonene is the main component. Limonene constitutes 40 to 67% of chemical compositions of Citrus EOs. Our result is consistent with the literature that limonene is the most abundant secondary metabolite in Citrus EOs [9]. The highest limonene level presents in Citrus spp. (67.76%), followed by C. limon (54.62%), C. nobilis (50.06%), C. amblycarpa (42.23%), and C. aurantifolia (40.84%).

Chemical Composition
Every EO has its chemical composition characteristics. C. aurantifolia EO contains the most diverse secondary metabolites (Table 2). Based on our results, the EO of C. aurantifolia is characterized by seventeen secondary metabolites, which are dominated by monoterpenes as many as twelve compounds. Additionally, the EO has secondary metabolites exclusive to C. aurantifolia. These consist of −elemene and (+)-4-carene monoterpenes, as well as a caryophyllene sesquiterpenoid. The EO of C. nobilis contains monoterpene and monoterpenoid secondary metabolites, which spread evenly ( Table 2). The EO has unique compounds: trans-carveol, citronellal, and geraniol. Like C. aurantifolia, C. limon produced EO with chemical composition dominated by monoterpenes. The EO contains three aromatic compounds, including o-cymene, p-cymene, and thymol. The two last compounds and α-pharnesene are exclusive to C. limon EO. Likewise, C. amblycarpa EO contains monoterpenes as the main types of secondary metabolites. The EO has a relatively high content of terpinen-4-ol and citronellol compared to the other Citrus EOs (Table 2). Meanwhile, Citrus spp. EO has the lowest diversity of secondary metabolites. Only four secondary metabolites constitute Citrus spp. EO. They are limonene, β-myrcene, −ocimene, limonene oxide, and citral. The levels of limonene and citral are the highest in Citrus spp. EO. Furthermore, limonene oxide is limited to this EO.

Antioxidant Activity
Antioxidant activities of Citrus EOs are represented by the percentage of free radical scavenging (%) ( Figure S2) and the concentration of 50% scavenging ( Table 2). Each EO exhibits a different inhibition profile in scavenging DPPH free radicals ( Figure S2). The EO of C. limon shows the highest antioxidant activity with IC50 of 4.25 ± 0.08 μL/mL (Table 2), consistent with the work published by Frassinetti and colleagues [10]. The antioxidant activity of C. limon EO even higher than that of ascorbic acid. In the second place, C. amblycarpa shows an IC50 value of 6.30 ± 0.04 μL/mL. The EOs of C. aurantifolia and Citrus spp. display similar calculated IC50 values, 12.85 ± 0.20 and 13.29 ± 0.04 μL/mL, respectively. Meanwhile, the EO of C. nobilis possesses the weakest antioxidant activity with an IC50 value of 15.83 ± 0.04 μL/mL. These different antioxidant activities of EOs may be due to their chemical composition characteristics.

Principal Component Analysis
As discussed above, we have identified twenty-five secondary metabolites, which are sparsely distributed among the EOs from five different Citrus species. Those secondary metabolite profiles may contain redundant information, which raises an issue in elucidating antioxidant activities of Citrus EOs. Therefore, we transformed Table 1 into a matrix (Table S1) and performed a PCA approach. Through linear combinations, PCA transformed correlated/uncorrelated descriptors, including twenty-five secondary metabolites. Subsequently, the essential information was extracted into only four uncorrelated variables or principal components (PCs) (Figures 1 and 2).
Thus, the fifteen secondary metabolites, which are shaded by gradient colors of dark blue in Figure  1, have good representations on PC1 and 2. Meanwhile, PC3 is important for (+)-4-carene, caryophyllene, and -elemene, with representation qualities of 71%. For -pinene and β-myrcene, the representation qualities of PC3 are 58 and 51%, respectively. According to Figure 2.a., the proportion of variance explained by each PC is not highly different. PC1 captured 33.8% of the total variance, while PC2 and 3 explain 26.9 and 23.1%, respectively. The rest variance is described as much as 16.3% by PC4. Combination of PC1 and 2 only explain 60.7% of the total variance (Figure 2.b). Thus, we may need at least three PCs to explain more than 80% of the total variance contributed by secondary metabolite descriptors. Intriguingly, the score plot on the coordinate system of PC1 and 2 revealed an informative result (Figure 3.a). In such a coordinate system, Citrus EOs spread according to their antioxidant activities. The EOs with IC50 above 12 ppm (Table S3) cluster in the third and fourth quadrants. On the other hand, the EOs with IC50 around 5 ppm are on the opposing side. The score plot (Figure 3.a) displays that the PC2 alone, which explains 26.9% of the total variance, can separate the essential oils based on their antioxidant activities. Nevertheless, the presence of PC1 assists in further separation of Citrus spp., C. nobilis, and C. aurantifolia. Furthermore, PC1 also helps the analysis through a biplot ( Figure  3.b).
The biplot (Figure 3.b) explains the difference between Citrus EOs based on their secondary metabolites. As an example, C. limon EO is characterized by thymol, o-cymene, p-cymene, αpharnesene, and -terpinene. Thymol, p-cymene, and α-pharnesene present only in C. limon EO, whereas the levels of o-cymene and -terpinene in the EO are the highest among other Citrus EO. For C. amblycarpa EO, camphene, terpinen-4-ol, -terpineol, 2-carene, -bisabolene, and citronellol describe the oil characteristic. In C. amblycarpa EO, the levels of those secondary metabolites are relatively higher than in other Citrus EOs. Interestingly, the biplot result supports the chemical composition characteristics of EOs in the sub-section on Chemical Composition. As discussed above regarding the score plot (Figure 3.a.), only PC2 can separate Citrus EOs on the factor map based on antioxidant activity ( Figure S1). Therefore, in the context of antioxidant activity, we only consider secondary metabolites that are highly represented by PC2 (Figure 1). These secondary metabolites are -terpinene, -ocimene, thymol, o-cymene, p-cymene, and α-pharnesene.
Not surprisingly, C. limon EO possesses the highest antioxidant activity among others since its character is determined by -terpinene, o-cymene, thymol, p-cymene, and α-pharnesene. The last three secondary metabolites present only in C. limon EO. Additionally, the levels of -terpinene  (Table S1 and S3).

Chemical Structures of the Antioxidants
Our study using PCA suggests that thymol, o-cymene, p-cymene, -ocimene, α-pharnesene, terpinene ( Figure 4) are the metabolites contributing to the antioxidant activities, particularly for C. limon EO. Thymol has antioxidant activity because it is a phenolic compound. Its hydroxyl group inhibits DPPH• radicals through hydrogen atom transfer (HAT). After transferring its hydrogen atom, thymol becomes a stable radical (thymol•) and reacts with other radicals to form DPPH-thymol or thymol-thymol [20].   [21]. The abstraction of the benzylic hydrogen atoms quenches free radicals, while the secondary metabolites turn into stable radical molecules. In both α-pharnesene and -ocimene, hydrogen atom transfer from carbon 5 to DPPH• molecules yield stable radicals. The stabilization comes from conjugated double bonds in α-pharnesene and ocimene structures, at carbon 1 and 3.
-Terpinene has antioxidant activity even it does not have conjugated double bonds [22]. It is a strong antioxidant similar to α-tocopherol, a well-known antioxidant [23]. Liu and Li [24] reported that -terpinene can directly scavenge free radical by donating its hydrogen atom at the allylic position, particularly carbon 3 and 6 ( Figure 6). Additionally, γ-terpinene has been reported to retard linoleic acid peroxidation [22]. Such activity undergoes through a rapid cross-reaction between hydroperoxyl (HOO•) and linoleylperoxyl (LOO•) which terminates free radical chain reaction. Moreover, the retardation of linoleic acid peroxidation converts γ-terpinene to p-cymene, an antioxidant.

Citrus Essential Oil Preparation and Gas Chromatography-Mass Spectrophotometry
Fruits of C. aurantifolia, C. nobilis, C. limon, C. amblycarpa, and Citrus spp. were collected from Caringin Central Market, Bandung, West Java, Indonesia during July to August 2018. The peels of C. aurantifolia, C. nobilis, C. limon, C. amblycarpa, and Citrus spp. were separately cleaned and cut into ± 2 cm. Each sample of Citrus peel (500 g) was subjected into a Clevenger-type hydrodistillation for 8 hours. The obtained EOs were dried using anhydrous sodium sulfate and stored in airtight containers at 4 ℃ before gas chromatography-mass spectrometry (GC-MS) analysis.

Gas Chromatography-Mass Spectrophotometry (GC-MS)
Chemical compositions of EOs were analyzed using GC-MS (Agilent GC Type 7890A amd MS Type 5975C). The analysis utilized DB 35MS column with length of 35 m, internal diameter of 0.25 mm, and film thickness of 0.25 μm. Nitrogen was used as the carrier gas at a flow rate of 1 mL/min. Every citrus EO was diluted in n-hexane to achieve concentration of 0.5% (v/v). Each one microliter EO solution was loaded to an injector with a temperature of 250 ℃. The column was initially held as 50 ℃ for 1 minute and increased to 250 ℃ with a heating ramp of 3.5 ℃/min.

Antioxidant Activity Assay
Antioxidant activities of Citrus EOs were determined using 2,2-diphenyl-2-picrylhydrazyl (DPPH) assay with a modification [25]. Briefly, 0.1 mL EOs at different concentrations (Table S3) were mixed with 2 mL DPPH (0.21 mM in ethanol 95%), in duplicate, and incubated in the dark for 60 min. After incubation, the absorbance was measured at 517 nm. As the control, ethanol was used instead of EO. Ascorbic acid was employed as the standard antioxidant compound. DPPH free radical scavenging activity was computed by utilizing the equation below: Acontrol denotes control absorbance, whereas Asample is sample absorbance. Antioxidant activity was also expressed as IC50 or concentration of EO (mg/mL) required to inhibit 50% DPPH. IC50 was determined from the plot % DPPH scavenging activity vs. EO concentration.

Principal Component Analysis
Peaks of secondary metabolites, from the GC-MS chromatogram of every citrus essential oil, were manually curated and assigned to a matrix (row i, column k) as listed in Table S1. The citrus species were treated as observation (i), whereas the secondary metabolites such as α-pinene and camphene were descriptors (k). Two types of preprocessing were performed to the matrix. The first one was mean centering to retain only the informative variation. The second one was scaling since levels of secondary metabolites in five citrus essential oils displaying different scales.
The matrix was subjected to an orthogonal linear transformation to generate a new coordinate system of principal components [26]. On the axis of rank s, (resp. ) indicates the coordinate vectors of the citrus species (resp. descriptors). They can be expressed as: ( ) represents the coordinate of each i on the axis s, whereas ( ) is the coordinate of descriptor k on the axis s. Notation s the eigenvalue corresponds to the axis s, the weight corresponds to descriptor k, is the weight corresponding to each i, is the general term of the matrix (row i, column k).