Variation in the Chemical Composition of Endemic Specimens of Hedychium coronarium J. Koenig from the Amazon and In Silico Investigation of the ADME/Tox Properties of the Major Compounds

Four species of the genus Hedychium can be found in Brazil. Hedychium coronarium is a species endemic to India and Brazil. In this paper, we collected six specimens of H. coronarium for evaluation of their volatile chemical profiles. For this, the essential oils of these specimens were extracted using hydrodistillation from plant samples collected in the state of Pará, Brazil, belonging to the Amazon region in the north of the country. Substance compounds were identified with GC/MS. The most abundant constituent identified in the rhizome and root oils was 1,8-cineole (rhizome: 35.0–66.1%; root: 19.6–20.8%). Leaf blade oil was rich in β-pinene (31.6%) and (E)-caryophyllene (31.6%). The results from this paper allow for greater knowledge about the volatile chemical profile of H. coronarium specimens, in addition to disseminating knowledge about the volatile compounds present in plant species in the Amazon region.


Introduction
Zingiberaceae has more than 50 genera and about 1400 plants [1]. The species that occur in Brazil are distributed in eight genera (Alpinia L., Amomum Roxb., Curcuma L., Etlingera Giseke, Hedychium coronarium J. Koenig, Kaempferia L., Renealmia L.f e Zingiber Boehm) and, among these, the genus Hedychium stands out for having the second largest number of representatives in Brazil (four species) [2].
The species Hedychium coronarium J. Koenig is endemic to India and China. In Brazil it is considered invasive and is popularly known as "lily of the marsh", "butterfly lily", "white-lily garland", "narcissus", "napoleon", "Olympia" or "white ginger", in addition to having several uses in folk medicine [3,4].
Hedychium species are cultivated as medicinal plants, ornamental plants, spices, and condiments [5]. The population of Malaysia uses this species for the treatment of gastric disorders, such as indigestion [6], and the medicinal drink is produced in the form of tea by infusion or decoction [7]. In Thailand, they use the tea from this plant, produced by infusion, for the treatment of osteoarthritis, caused by the wear and tear of cartilage in the joints, in addition to using the tea produced with the stem to treat tonsillitis. The rhizome of the plant has several medicinal properties, being used in Thailand to combat the excessive amount of gases produced after digestion [8]. In Vietnam it is used as a healing treatment and an antiseptic, fighting bacteria in the case of wounds. In Brazil it is consumed as a diuretic or to treat hypertension [9]. EOs extracted from the rhizome of H. coronarium have shown anthelmintic activities [10,11] and antimicrobial activities [12], in addition to having a phytotoxic effect [13]. For the leaves and also the rhizomes, fibrinogenolytic, coagulant [14], larvicidal [15], and antioxidant [16][17][18] activities have been found.
In Hawaii, the flowers are eaten as vegetables and used as garlands, and in Japan, they are used for perfume production.
Considering that EOs can lead to the discovery of new chemotypes of a species and that these variations are influenced by geographic and environmental factors [22][23][24][25], the present work aims to evaluate the chemical composition of the EOs for six specimens of H. coronarium collected in different municipalities of the state of Pará, Brazil. We aim to provide valuable scientific data to enhance the understanding of volatile compound profiles in the Amazon region. Furthermore, our data significantly support the comprehension of new specimens of H. coronarium. Through a comprehensive and rigorous approach, we innovatively investigated the diversity and composition of volatile compounds present in this unique region. Tables 1 and 2 show the 65 chemical compounds found in the EOs of the six specimens. The EO of specimen A, collected on the border between the states of Pará-Maranhão, Brazil, was characterized by 1,8-cineole (66.1%) and β-pinene (21.9%) in the rhizomes; on the other hand, the leaves from this species showed a majority of β-pinene (48.9%) and 1,8-cineole (66.1%). A majority of 1,8-cineole (46.2%) and β-pinene (31.1%) characterized, respectively, the EOs of the rhizomes and leaves from specimen B, collected in the municipality of Tracuateua, Pará, Brazil. In the EO of sample C, collected in the municipality of Igarapé-Miri, Pará, Brazil, 1,8-cineole (37.4%) in the rhizomes and β-pinene (34.8%) in the leaves were found in the majority. In specimen D, collected in Santarém Novo, Pará, Brazil, the EO was characterized by 1,8-cineole (35%) in the rhizomes and β-pinene (31.6%) in the leaves. In Belém, Pará, Brazil, two samples were collected (E and F), the EO of sample E presented the majority as β-pinene in the rhizomes (30.5%) and in the leaves (41%), while sample F presented 1,8-cineole (33.5%) in rhizomes and α-pinene in the leaves (32.9%).
β-pinene is one of the major compounds that showed significant significance in the results. This hydrocarbon monoterpene is found in the EOs of many coniferous plants, such as pine (Araucaria angustifolia) [38], and some research involving EOs has shown that this compound has antioxidant properties, biological activities against bacteria (Acetobacter calcoacetica, Bacillus subtilis, Clostridium sporogenes, Clostridium perfringens, Escherichia coli, Salmonella typhi, Staphylococcus aureus, and Yersinia enterocolitica) and fungi (Candida albicans, Aspergillus niger, Aspergillus flavus and Penicillium notatum) [39]. In turn, α-pinene is described in the literature as having a modulating action of antibiotic resistance against the multidrug-resistant bacterium Campylobacter jejuni that causes gastroenteritis [40,41], and studies report that α-pinene has a greater antimalarial property than β-pinene [41].
Regarding the specific question on whether the chemotypes of the leaves and rhizomes were the same for the same sample, our research indicates that there are variations in the chemical composition of both plant parts within a single specimen. The differences observed between the leaves and rhizomes highlight the importance of considering different plant organs when studying volatile compounds in H. coronarium. These findings suggest that the biosynthesis and accumulation of volatile compounds may be organ specific, indicating potential variations in their ecological roles and chemical profiles.

Multivariate Analysis
The chemical compounds were identified in the different fractions of EOs of H. coronarium. The multivariate analysis PCA (principal component analysis) is shown in Figure 1 and the HCA (hierarchical cluster analysis) is shown in Figure 2. In Figure 1, we can see that PC1 explains 49.4%, while PC2 explains 27.3% of the variations, and the two components add up to 76.7% of the variance. When analyzing the HCA, considering the Euclidean distances and complete bonds (Figure 2), we have the formation of three distinct groups formed by the fractions, with group I formed only by sample A1, while group II is formed by samples B1, E1, C1, and D1, with a similarity of 41.01% (Figure 2), while group III did not show a significant level of similarity with any sample rhizome EO.
A multivariate analysis was applied to analyze the similarity in the chemical composition between the different fractions of EOs isolated from the H. coronarium leaves. Figure 3 shows the principal component analysis (PCA), while Figure 4 shows the hierarchical cluster analysis (HCA), according to which we can observe with the results obtained in the PCA (Figure 3) that the first component explains 36.9%, while PC2 explains 31.9% of the variances, the sum of the two components explains 68.8% of the variations observed in Figure 4 of the HCA. We note that there was the formation of three groups, the first group was formed by the samples of oils A2 and E2, group II was formed by the samples B2, C2, and D2, while group III was formed only by the sample F2, with a similarity of 26% of samples A2 and B2, 36.59% between samples B2, C2, and D2, and 5.11% of sample III in relation to sample II, that is, a low similarity between them ( Figure 4). In addition, in Figure 3, it is possible to observe which compounds were responsible for positively or negatively impacting the formation groups, for example in group I the highest number of compounds were 1,8-cineole, β-pinene, and camphor, in group II they were 1-epi-cubenol, α-terpineol, sabinene, α-muurolol, spathulenol, epi-a-muurolol, caryophyllene oxide, epi-α-cadinol, terpi-ne-4-ol, humulene epoxide II, and (E)-caryophyllene, while in group III they were limonene y-terpinene, (E)-β-cimene, α-humulene, myrcene, and α-pinene. Chemometric analysis has been shown to be an important tool for researchers of natural products, because through it they can analyze the differences between samples of EOs using matrix correlation, which shows the differences and similarities between samples of different plants or samples collected at different times, or different regions of ions [53].  In addition, Figure 1 shows the compounds that each characteristic group formed in the multivariate analysis, for example, group I, which comprises the largest number of grouped samples, was formed by the compounds 1,8-cineole and linalool. On the other hand, in group II, the compounds that contributed positively to the similarity between the different fractions were p-cymene, sabinene limonene, camphene, (E)-sabinene hydrate, d-elemene, a-thujene, b-pinene, and camphor, and in group III α-terpineol, (E)-caryophyllene, borneol α-terpinyl acetate γ-terpinene, α-phellandrene, terpinolene, terpinen-4ol, myrcene, α-terpinene, and α-pinene ( Figure 1).  Multivariate analysis was used to verify the potential similarity of the different fractions of EOs obtained from the vegetative organs, rhizomes, and leaves of H. coronarium. In addition, we can see in the graph that each component explains a value of the variance in the analyzed data, for example, the first component explains 40% and the second component explains 22% of the variance ( Figure 5). In the HCA hierarchy analysis, Figure 6, we can analyze the formation of the different groups. In general, there was the formation of four groups, with different degrees of similarity. Group I, with a similarity of 54.03%, was formed by the samples A1 and A2, essential oils from the rhizome and leaves, respectively. Group II was formed by samples of the EO only isolated from the rhizome, namely B1, E1, C1, and D1, with a similarity of 59.64%. Group III, with a similarity of 39.26%, was formed only by a sample of essential oils isolated from the leaves, namely B2, D2, E2, and F2. The sample F1 group IV followed the same pattern already described in Figure 2, that is, it had no similarity with the other samples. These results demonstrate that plant organs can biosynthesize different substances in qualitative and quantitative terms. A multivariate analysis was applied to analyze the similarity in the chemical composition between the different fractions of EOs isolated from the H. coronarium leaves. Figure  3 shows the principal component analysis (PCA), while Figure 4 shows the hierarchical cluster analysis (HCA), according to which we can observe with the results obtained in the PCA (Figure 3) that the first component explains 36.9%, while PC2 explains 31.9% of the variances, the sum of the two components explains 68.8% of the variations observed in Figure 4 of the HCA. We note that there was the formation of three groups, the first group was formed by the samples of oils A2 and E2, group II was formed by the samples B2, C2, and D2, while group III was formed only by the sample F2, with a similarity of 26% of samples A2 and B2, 36.59% between samples B2, C2, and D2, and 5.11% of sample III in relation to sample II, that is, a low similarity between them ( Figure 4). In addition, in Figure 3, it is possible to observe which compounds were responsible for positively or negatively impacting the formation groups, for example in group I the highest number of compounds were 1,8-cineole, β-pinene, and camphor, in group II they were 1-epi-cubenol, α-terpineol, sabinene, α-muurolol, spathulenol, epi-a-muurolol, caryophyllene oxide, epiα-cadinol, terpi-ne-4-ol, humulene epoxide II, and (E)-caryophyllene, while in group III they were limonene y-terpinene, (E)-β-cimene, α-humulene, myrcene, and α-pinene. Chemometric analysis has been shown to be an important tool for researchers of natural products, because through it they can analyze the differences between samples of EOs using matrix correlation, which shows the differences and similarities between samples of different plants or samples collected at different times, or different regions of ions [53].   The multivariate PCA analyses were carried out in the factorial plane for the samples of essential oils from the leaves and rhizomes. In the PCA, we can observe that PC1 explains 71.9% of the variance and PC2 explains 20.3%. In Figure 7, it is possible to analyze that three oil samples are separated from F2, A1, and A2, corroborating the previous results of the chemometric analysis for the compounds, Figures 3 and 5. In addition, a HCA hierarchy analysis (Figure 8) was carried out, the results of which corroborate those presented in all the previous graphs; for example, the compounds that had the highest weights for the formation of the groups, the group was formed only by the F2 sample, this may be related to the presence of α-pinene in a higher concentration. Group II was formed by the other samples (B1, C1, D1, E1, F1, A2, B2, C2, D2, and E2), and the relationship between them is in the presence of the compounds sabinene, β-pinene, myrcene, α-phellandrene, α -terpinene, p-cymene, Limonene, 1, 8-cineole, (E)-β-ocimene, γ-terpinene, terpinolene, linalool, Camphor, borneol, terpinen-4-ol, p-cymen-8-ol, and α-terpineol. Group III and IV are formed by separate samples A2 and A1, with the most representative compounds β-pinene and 1,8-cineole, respectively, being in agreement with the HCA analysis, as shown in Figures 4 and 6.  Multivariate analysis was used to verify the potential similarity of the different fractions of EOs obtained from the vegetative organs, rhizomes, and leaves of H. coronarium.  In addition, we can see in the graph that each component explains a value of the variance in the analyzed data, for example, the first component explains 40% and the second component explains 22% of the variance ( Figure 5). In the HCA hierarchy analysis, Figure 6, we can analyze the formation of the different groups. In general, there was the formation of four groups, with different degrees of similarity. Group I, with a similarity of 54.03%, was formed by the samples A1 and A2, essential oils from the rhizome and leaves, respectively. Group II was formed by samples of the EO only isolated from the rhizome, namely B1, E1, C1, and D1, with a similarity of 59.64%. Group III, with a similarity of 39.26%, was formed only by a sample of essential oils isolated from the leaves, namely B2, D2, E2, and F2. The sample F1 group IV followed the same pattern already described in Figure 2, that is, it had no similarity with the other samples. These results demonstrate that plant organs can biosynthesize different substances in qualitative and quantitative terms.  The multivariate PCA analyses were carried out in the factorial plane for the samples of essential oils from the leaves and rhizomes. In the PCA, we can observe that PC1 explains 71.9% of the variance and PC2 explains 20.3%. In Figure 7, it is possible to analyze that three oil samples are separated from F2, A1, and A2, corroborating the previous results of the chemometric analysis for the compounds, Figures 3 and 5. In addition, a HCA hierarchy analysis (Figure 8) was carried out, the results of which corroborate those presented in all the previous graphs; for example, the compounds that had the highest weights for the formation of the groups, the group was formed only by the F2 sample, this may be related to the presence of α-pinene in a higher concentration. Group II was formed by the other samples (B1, C1, D1, E1, F1, A2, B2, C2, D2, and E2), and the relationship between them is in the presence of the compounds sabinene, β-pinene, myrcene, α-phellandrene, α -terpinene, p-cymene, Limonene, 1, 8-cineole, (E)-β-ocimene, γ-terpinene, terpinolene, linalool, Camphor, borneol, terpinen-4-ol, p-cymen-8-ol, and α-terpineol. Group III and IV are formed by separate samples A2 and A1, with the most representative compounds β-pinene and 1,8-cineole, respectively, being in agreement with the HCA analysis, as shown in Figures 4 and 6.

In Silico ADMET Analysis
Due to the limited pharmacokinetics and metabolic performance of essential oils, they often fail to meet the requirements for antimicrobial/antibacterial drug testing [54][55][56]. Therefore, we conducted an analysis of the ADMET profile for the main constituents found in the tested essential oils. Our analysis retained the calculations of more than 50 ADMET parameters for the studied compounds, namely 1,8-Cineole, α-Pinene, β-pinene, and (E)-caryophyllene. Table 3 provides an overview of the estimated ADMET properties for the selected compounds. Lipinski's Rule of Five, introduced by Dr. Christopher Lipinski, is a guideline in drug design. It assesses a compound's oral bioavailability based on its molecular weight, lipophilicity, hydrogen bond donors, and acceptors. These criteria help determine a compound's drug-likeness and potential for successful oral administration. In accordance with important drug-likeness guidelines, a compound should not violate more than one Lipinski rule. Furthermore, its molecular weight should be below 500 g/mol, its topological surface area (TPSA) should be less than 140 Å 2 , the number of H-bond acceptors (nOHA) should not exceed five, the number of H-bond donors (nOHD) should be five or less, the water partition coefficient (WLOGP) should not exceed 5.88, and the number of rotatable bonds (nRB) should be ten or less [57,58]. As per Table 3, those compounds violating more than one parameter would be considered as a Lipinski violation. Based on our findings, all the compounds had a TPAS less than 30 Å 2 . Except for α-pinene, β-pinene, and (E)caryophyllene, all the compounds exhibited high gastrointestinal absorption (GI), indicating their easy absorption through the gastrointestinal tract. Many of the compounds were found to be (theoretically) soluble in water (except terpenes and sesquiterpenes), which is an important criterion for their effectiveness as a drug. One of the major components of the EO, α-pinene is a colorless, water-insoluble, but oil-and ethanol-soluble organic liquid. β-pinene is also a colorless organic liquid, which is oil soluble but ethanol-and water-insoluble. Moreover, 1,8-cineole is insoluble in water, 3.50 × 10 3 mg/L at 21 • C, and also miscible with ether, alcohol, chloroform, glacial acetic acid, and oils. As per the data available from the National Institutes of Health (NTP), 1992, it is insoluble at <1 mg/mL at 68 • F. However, data published by the Joint FAO/WHO Expert Committee on Food Additives (JECFA) states that it is insoluble in water and miscible in oils. However, there is different information about this on the PubChem website (https://pubchem.ncbi.nlm.nih. gov/compound/Eucalyptol#section=Solubility, accessed on 1 July 2023). (E)-caryophyllene is soluble in ether and ethanol, and insoluble in water. During the absorption process, first-pass metabolism via P-glycoprotein (P-gp) and cytochrome P450 enzymes in the small intestine and liver can negatively impact drug bioavailability. However, our results indicated no P-glycoprotein (P-gp) substrates among the compounds, suggesting good intestinal absorption, while some compounds mainly interacted with two isoenzymes of the cytochrome (CYP450) family, specifically CYP2C19 and CYP2C9, indicating their effectiveness with minimal toxicity. (E)-caryophyllene was predicted to be unable to cross the blood-brain barrier (BBB), as shown in the boiled-egg prediction. Compounds located in the yellow zone of the graph can permeate through the blood-brain barrier (BBB). The drug-like properties and gastrointestinal (GI) absorption of the chosen compounds from the essential oils were assessed using the boiled-egg prediction ( Figure 9) and bioavailability radar graph ( Figure 10). Compounds located in the yellow zone of the boiled-egg graph have the ability to cross the blood-brain barrier (BBB), while the pink area on the bioavailability radar graph indicates their drug-like characteristics.

Material
Samples

Preparation of the Botanical Material
The A-F samples of H. coronarium leaves were dried in an oven with air circulation at 35 °C for five days and then ground in a knife mill (Tecnal, model TE-631/3, Piracicaba, São Paulo, Brazil).

Material
Samples

Preparation of the Botanical Material
The A-F samples of H. coronarium leaves were dried in an oven with air circulation at 35 °C for five days and then ground in a knife mill (Tecnal, model TE-631/3, Piracicaba, São Paulo, Brazil). Additionally, the toxicological properties of the compounds were assessed and presented in Table 3. None of the selected compounds exhibited organ or oral toxicity, except for (E)-caryophyllene. In summary, based on the results, it can be concluded that these compounds have the potential for further development as drug candidates. The LD 50 values were also calculated to ensure the safety of the selected compounds, as shown in Table 3. The compounds with LD 50 > 2000 mg/kg suggest their safety for biological administration and as potential drugs.

Extraction of Volatile Compounds
Our in silico results for the major compounds of different H. coronarium essential oils match with earlier reported data [59].

Material
Samples A-F of H. coronarium were collected in the state of Pará: Sample A (highway Pará-Maranhão Km 290), Sample B (municipality of Tracuateua), Sample C (municipality of Igarapé-Miri), Sample D (municipality of Santarém Novo), Samples E and F (Belém). Voucher specimens were deposited in the herbarium at the Museu Paraense Emílio Goeldi (Sample B: MG182,830, Sample E: MG182,843, and Sample F: MG177,796). The other samples were identified by comparison with authentic voucher plants.

Preparation of the Botanical Material
The A-F samples of H. coronarium leaves were dried in an oven with air circulation at 35 • C for five days and then ground in a knife mill (Tecnal, model TE-631/3, Piracicaba, São Paulo, Brazil).

Extraction of Volatile Compounds
The samples were subjected to hydrodistillation in modified Clevenger-type glass systems for 3 h, coupled with a refrigeration system to maintain the condensation water at around 12 • C, following protocols reported earlier by our research group [4,60].

Analysis of the Volatiles
The phytochemical profiles of the EOs were analyzed using chromatography/mass spectrometry (GC/MS) using a Shimadzu QP Plus 2010 GC-MS (Kyoto, Japan), following protocols reported earlier by our research group [4,60]. The retention index was calculated for all the volatile constituents using a homologous series of n-alkanes (C8-C40, Sigma-Aldrich, St. Louis, MO, USA), according to Van den Dool and Kratz [61], and the compounds were identified by comparing their mass spectrum and retention index with the data from the libraries [52].

ADMET Analyses
In modern drug-like hit identification, estimations of the pharmacokinetic properties have a crucial role in it [62,63]. Nowadays, many machine learning-based theoretical ADMET analyses tools are available online, which helps scientists to get more insights from these properties before actually going for higher pre-clinical studies. Although they have their own limitations, certainly the tools with good applicability domains have higher chances of accurate predictions. 'SwissADME' is one of the tools available online, which is useful for theoretical ADMET assessments [64,65]. One important mechanism that underpins drug-drug interactions is the induction or inhibition of CYP enzymes. Considering this fact, our in silico analyses for CYP1A2 inhibition, CYP2C19 inhibition, CYP2C9 inhibition, CYP2C9 substrate, CYP2D6 inhibition, CYP2D6 substrate, and CYP3A4 inhibition profiles were retained negatives. This also suggested that these EO components can be used further or modified accordingly to more suitable derivatives in order to have more drug-like candidates.
The chemical structures of the chosen compounds from the essential oils were drawn first using the ChemDraw Ultra 8.0 software for the purpose of investigating their theoretical pharmacokinetics, which includes absorption, distribution, metabolism, and excretion (ADME). The accompanying descriptions were converted into the SMILES format. To assess the drug-like and pharmacokinetic characteristics of the selected compounds, we utilized the ADME tool provided by the SwissADME online server (http://www.swissadme.ch/, accessed on 1 June 2023), following a predefined procedure. To evaluate their toxicity profile, we employed the ProTox-II webserver (http://tox.charite.de/protox_II, accessed on 1 June 2023). This server utilizes various parameters, such as organ toxicity (hepatotoxicity), oral toxicity, and toxicological endpoints (cytotoxicity, mutagenicity, carcinogenicity, and immunotoxicity), to make predictions. From our analyses of the EO components using this tool we noted down important ADMET properties (Table 3).

Statistical Analysis
Multivariate analysis was performed according to the methodology described by [27], where the Minitab 17 ® software (free version, Minitab Inc., State College, PA, USA) was used.

Conclusions
This paper investigated the chemical composition of the essential oils from six endemic specimens of H. coronarium in the Amazon region. The identification of variations in the chemical composition of H. coronarium essential oils contributes to the exploration of the species' ecological and evolutionary aspects. By understanding how geographic and environmental factors shape the volatile profiles, we can gain insights into the adaptive mechanisms of H. coronarium and its interactions within its natural habitat. By uncovering the chemical diversity and complexity within this species, we open new avenues for the discovery of potential bioactive compounds and novel applications in various industries. The use of multivariate analysis enabled us to monitor the variability of the volatile compounds, both in terms of the compound classes and individual compounds, through the construction of a correlation matrix. This analytical approach allowed us to identify distinct chemotypes among the specimens studied, highlighting the intricate nature of H. coronarium volatile composition. The utilization of multivariate analysis techniques, along with the consideration of different plant organs, allowed us to unveil the complex variations and chemotypes present within this species. These results not only contribute to our knowledge of volatile compound profiles in the Amazon, but also have broader implications for ecological, pharmaceutical, and agricultural research. Overall, our findings advance the frontiers of knowledge in this field and lay the groundwork for future investigations into the chemical diversity and ecological significance of H. coronarium. In conclusion, our study contributes to a deeper understanding of the volatile compound profiles within the Amazon region and sheds light on the chemical diversity present in six endemic specimens of H. coronarium.