Helichrysum italicum (Roth) G. Don Essential Oil from Serbia: Chemical Composition, Classification and Biological Activity—May It Be a Suitable New Crop for Serbia?

H. italicum essential oil (EO) is one of the most popular ingredients utilized by the cosmetic industry, and it is also used as natural antioxidant and as a value-added ingredient in food products. The chemical composition of the EO H. italicum cultivated in Serbia was analyzed by gas chromatography–mass spectrometry. The quantitative structure–retention relationship was used to predict the retention indices of the EO constituents acquired by GC-MS data, applying five molecular descriptors selected by factor analysis and a genetic algorithm. Also, antimicrobial activity, and biological activity by four common antioxidant tests (DPPH and ABTS assays, reducing power, and β-carotene bleaching test), and in vitro antihyperglycemic and anti-inflammatory capacities were evaluated. A total of 70 EO constituents were detected, of which 17 (8.5%) could not be identified. The H. italicum EO in this study belonged to γ-curcumene chemotype. The coefficients of determination reached the value of 0.964, demonstrating that this model could be used for prediction purposes. All applied tests showed that H. italicum EO possesses good biological activity and an interesting chemical composition. Therefore, the EO of H. italicum grown in Serbia has a potential to be used in food, cosmetic, and pharmaceutical products.


Introduction
The genus HelichrysumMill. includes over 300 species out of which 25 are distributed in Europe and the Mediterranean, while the others are distributed in Africa and Madagascar, Australia, and New Zealand [1][2][3], and also in temperate regions of Asia, including in India, Iran, and Turkey [4][5][6]. Mediterranean species belong to the Stoechadina section [7], and H. italicum (Roth) G. Don (syn. H. angustifolium D.C.) is the most widespread species, and it is found on alkaline, dry, sandy and poor soil [8] or even on coastal rocks where it identifies habitats protected by the European directive [9,10]. However, it is characterized by a high degree of anatomical and morphological polymorphism traits and a diversity of ecotypes [11,12]. Some botanists think that H. italicum is a complex species, which includes three species: H. litoreum Guss., H. serotinum Boiss. with two subspecies: ssp.  [8,13].
Today, immortelle or H. italicum essential oil (EO) is one of the most popular ingredients in cosmetic products, especially in skin care products, as it exhibits antiproliferative and tissue remodeling effects, thus helping with the wound healing process. Furthermore, as this EO stimulates blood circulation in the skin, it enables regeneration and has antiaging effects [14,15]. Essential oil composition depends on population, altitude, and climatic conditions [16], developmental stage (before flowering, full blossom, after flowering) [17,18], ecology and plant communities [19], plant part (leaves or flowers) [20], extraction method [20,21], etc.
H. arenarium is the only species in Serbian flora that grows spontaneously on sandy soil in dry grassy areas, mainly in the Vojvodina Province. Another species, H. italicum, is better known and cultivated to produce EO and extracts for the cosmetic industry. This plant is relatively new in agriculture as only wild populations were harvested in the past [8]. The growing interest in H. italicum can be attributed to the high price of its EO, especially in some countries in southern Europe, including the Balkans, Spain, and France [22]. Characteristic properties of EOs, as well as their quality and price, depend on their chemical composition [23]. Variability of the H. italicum EO composition is influenced by several factors including not only the plant growth stage and genotype, but also the geographic origin and environment [8]. Previous research showed that H. italicum EO exhibited antioxidant, antimicrobial, antiviral, anti-inflammatory, and antiproliferative activity [8].
The aim of this investigation was to characterize EO obtained by hydrodistillation from flowering parts of H. italicum introduced fromČapljina (Bosnia and Hercegovina) and grown in the experimental fields of the Institute of Field and Vegetable Crops in Novi Sad, Serbia. We conducted a systematic literature review of H. italicum EO composition using Science Direct Elsevier, SpringerLink, PubMed, Scopus, Scifnder, Web of Science, Wiley Online and Google Scholar, among others. The obtained data were used for cluster analysis and chemotyping of H. italicum. The main goal was to establish the quantitative structureretention relationship (QSRR) model for anticipating the retention indices (RIs) of certain compounds in H. italicum EO obtained by GC-MS chromatography. Also, antimicrobial, antioxidant, in vitro antihyperglycemic, and anti-inflammatory activities of H. italicum EO were tested in this study.

EO Extraction and Analysis
After drying, the plant material was fragmented and the EO was extracted by hydro distillation in Clevenger apparatus, dried over anhydrous sodium sulfate, and stored in vials at 4-6 • C. The EO yield was 0.17% (v/w). The GC-MS analysis was carried out using an Agilent 7890A apparatus equipped with a 5975 C MSD, FID and a HP-5MS fused-silica capillary column. The EO constituents were identified based on their linear retention index relative to C8-C32 n-alkanes, compared with data reported in the literature (Adams4 and NIST11 databases). The relative percentage of the oil constituents was expressed as percentages by FID peak area normalization.

Phylogenetic Tree Diagram
The phylogenetic tree diagram was plotted using R software 4.0.2 (64-bit version). This diagram was calculated using R package "ape" (Analysis of Phylogentics and Evolution), applied as a graphical tool to represent the arrangements of similar EOs concentration (evaluated in the cluster analysis). The obtained experimental results were presented in the matrix, after which the hierarchical cluster analysis was performed. The distance matrix was determined using euclidean method, while the cluster analysis was performed using the "complete" method.

QSRR Analysis
The determination of molecular descriptors (MDs) was done using the PaDel-descriptor software [24,25]. The most relevant MDs for RIs prediction by factor analysis and genetic algorithm (GA) [26,27], using Heuristic Lab software. Statistical investigation of the data was investigated by the Statistica 10 software.

Artificial Neural Network (ANN)
Multilayer perceptron architecture (MLP) was used to build the artificial neural network model (ANN) for prediction of RIs for compounds found in H. italicum EO identified using GC-MS data [28]. Broyde-Fletcher-Goldfarb-Shanno (BFGS) algorithm was used to speedup the calculation of weight coefficients of the ANN [29]. More details regarding the descriptors could be studied in the Handbook of Molecular Descriptors [30]. The observed data were randomly separated to 70%, 15%, and 15% of data used for training, testing, and validations, respectively [31,32]. ANN calculations were executed with Statistica10.

Global Sensitivity Analysis
Yoon's global sensitivity method was used to calculate the relative impact of the selected MDs on RIs, according to weight coefficients of the developed ANN [33].

Biological Activity
In this study, all biological testswere conducted in vitro. In case of antimicrobial tests, it was repeated several times until the same value was obtained three times. Antioxidant, antihyperglycemic, and anti-inflammatory tests were conducted in three replications, and values are expressed as mean ± standard deviation (SD).

Antimicrobial Activity
The antimicrobial activity of the H. italicum EO was evaluated using 16 strains of American Type Culture Collection, using microtitre plate-based antibacterial assay incorporating resazurin as an indicator of cell growth. As a control, a Gentamicin strip test was used for determination of minimal inhibitory concentration (MIC) [34].

In Vitro α-Glucosidase Inhibitory Potential
α-Glucosidase inhibitory potential was performed to examine in vitro antihyperglycemic activity (AHgA) of H. italicum EO using the method reported by Tumbas Šaponjac et al. (2014) [39], where each well contained 100 µL of 2 mmol L−14-nitrophenyl α-D-glucopyranoside in 10 mmol/L potassium phosphate buffer (pH 7.0) and 20 µL of the samples at concentration 250 mg/mL, diluted in buffer. The reaction was started with the addition of 100 µL of the enzyme solution (56.66 mU/mL). The plates were incubated at 37 • C for 10 min. The absorbance of 4-nitrophenol released from 4-nitrophenyl α-D-glucopyranoside was measured at 405 nm.

Anti-Inflammatory Activity
The anti-inflammatory activity (AIA) was determined by protein denaturation bioassay using egg albumin (from fresh hen's egg), according to method adopted by Ullah et al. (2014) [40], where the absorbance was measured at 660 nm.

Statistical Analysis
The results represented are means ± standard deviation. Statistical study of the data was done using the Statistica 10 software. The cluster analysis (CA) was employed to investigate intra-and interpopulation variations and distinctions of EO constituents of H. italicum EO in samples gathered at various locations and/or introduced from literature reports, where the data were evaluated utilizing R software, 4.0.2 (64-bit version).

Results
The chemical composition of the H. italicum EO in this study was determined by GC-FID and GC-MS analyses, as shown in Table 1. A total of 70 compounds were detected, among which 17 were unidentified and accounted for 8.5%. The dominant compounds were: γ-curcumene (13.6%), β-selinene (12.2%), and α-pinene (11.8%), followed by βcaryophyllene (6.7%), ar-curcumene (5.0%), α-selinene (4.3%), and italicene (4.2%). With the intention to develop a QSRR model for prediction of RIs, PaDel-descriptor software was used. A large set of molecular descriptors (MDs) were calculated, and only the highly important descriptors were selected to build the forecasting RIs model. According to this initial examination, only around 250 descriptors remained for genetic algorithm (GA) calculation. The GA was applied to select among MDs for the highly influential variables for RIs anticipation. As an outcome, the five highly important molecular descriptors selected by GA were: three autocorrelation descriptors (AATS1m, AATSC4c, and MATS1c), Barysz matrix descriptor (VE3_Dzv), and Van der Waals volume descriptor (VABC).
Subsequently, the used MDs were appropriate to foresee the RIs of compounds in H. italicum by multivariate artificial neural network (ANN) model. Proposed descriptors describe discrete aspects of the molecular bindings and were used to develop the QSRR models. The correlations between these molecular descriptors are presented in Table 2. Table 2. Correlations between molecular descriptors for H. italicum essential oil.
With an idea to investigate the nonlinear relationship among RIs of compounds in and MDs selected by GA, ANN modelling tool was used to build a predictive model. The ANN model MLP 5-7-1 was constructed to predict the retention indices of compounds isolated from H. italicum EO. The coefficients of determination during the training cycle was 0.997, indicating that this model could be applied for prediction of RIs, having in mind the low prediction error and high r 2 . The statistical results of the ANN model are shown in Table 3. The predicted RIs are presented in Figure 1a, confirming the good fit of the constructed ANN by explaining the relationship among the predicted and experimental RIs values. The obtained results presented in Figure 1b show the good quality of the ANN model for anticipating the RIs of compounds in H. italicum EO obtained by GC-MS analysis. In this chapter, the impact of five most influential input variables on RIs, selected using genetic algorithm, was investigated. Based on the results presented on Figure 1b, MATS1c was the most influential MD for chemical compounds in H. italicum EO, with relative importance of +53.84%. The positive influence was observed for VE3_Dzv and AATSC4c descriptors, which expressed the relative importance of +27.63 and +10.17%. The negative influential MD for H. italicum EO was: VABC (with relative importance of −7.94%).
H. italicum showed low or no activity against tested bacteria. However, for all Gram negative bacteria (E. coli, P. aeruginosa, Salmonella Typhimurium, S. enteritidis, K. aerogenes, and P. hauseri) MIC and MBC values were higher than 454.50 µL/mL EO. For the Gram positive bacteria (B. cereus, L. monocytogenes, R. equi, and S. epidermidis) MIC and MBC values was 454.50 µL/mL, while for other (B. spizizenii, E. faecalis, L. innocua, L.ivanovii, and S. aureus) MIC and MBC values were higher than 454.50 µL/mL of EO (as illustrated in Table 4).   Table 4). Furthermore, H. italicum EO at the concentration 250 mg/mL had strong inhibitory activity on α-glucosidase (66.02%) and inhibited the denaturation of the proteins by 37.41% (as illustrated in Table 4).
Samples with high content of α-pinene are distributed in Italy [42], Portugal [59], and Croatia [52], while moderate α-pinene content in H. italicum EOs was recorded in Bulgaria, where it was introduced by the Corsican population [60], Croatia [52], Bosnia and Hercegovina [47], and Algeria [48]. This indicates that the chemical composition varies depending on the weather conditions and altitude [16].
Juniper camphor is reported as the main compound for other plant species (Syzygium samarangense (Blume) Merr. & L.M.Perry, Artemisia argyi Lév. & Vaniot, Pulicaria somalensis Hoffm.), as an EO compound responsible for antimicrobial, antioxidant, and phytotoxic properties [61]. However, H. italicum with juniper camphor as the dominant compound was reported in Sardinia [16]. Samples with the dominant β-selinene in the EO were also reported in Italy (Potenza, Taranto, Bari, Matera, Basilicata, and Siena) [57]. Furthermore, β-selinene is an important constituent for perfume industry, and thus, is commercially significant [62]. Furthermore, the H. italicum EO from Italy (Tuscany) could be divided into two main groups based on its chemical composition: (1) rich in β-diketones or italidiones (with 32.8-42.0%); (2) rich in neryl acetate (39.9-44.5%) that contained between 3.7 and 9.7% of β-diketones [44]. Neryl acetate is characterized by an orange blossom, rose, sweet, and fruity odor [63], while ar-curcumene are the prime contributors to the characteristic 'ginger' attribute [64]. However, γ-curcumene, as a valuable antioxidant compound, is very unstable and is easily transformed into italicene and isoitalicene, and further into α-curcumene when exposed to light [65]. Chemotypes containing these compounds as the dominant ones in the EO could be used in fragrance and perfume industry. They can also be used as natural antioxidants in preventing deterioration of foodstuff, beverage products and pharmaceuticals [46]. Italidiones act as anti-inflammatory agents and protect the skin against pollution and UV radiation [66]. Compounds such as α-pinene showed inhibitory activity on both collagenase and elastase enzymes associated with skin aging process, and thus, it is a valuable raw material in the cosmetic industry [53].
However, some samples, such as the two from Sardinia [16] with cis-β-guaiene (58.2%) or trans-nerolidol (55.6%) as the dominant compounds, did not fit into any of the above mentioned chemotypes. This was also the case with the one sample from USA with neryl acetone as the dominant compound (38.6%) [49]. This could be a consequence of the environmental conditions or crossing with other species or hybrids.
Antimicrobial agents from natural origin in the recent years are extensive studies. However, it depends on strong influence of the regional origin on chemical composition investigated plant species and microbial strain [67]. Antimicrobial activity of H. italicum EO from Algeria with α-cedrene, ar-curcumene, and geranyl acetate as dominant compounds assayed by disk diffusion method inhibited growth S. aureus, M. luteus, E. cereus, B. cereus, S. epidermidis, B. subtilis, P. aeruginosa, E. faecalis, and P. mirabilis, but did not affect E. coli, K. pneumonia, and L. monocitogenes. In addition, yeasts (C. albicans and S. cervisae) as well as fungi (F. solani, A. niger, A. alternata and A. rabiei) were also inhibited by H. italicum EO [66].
Antioxidants are mainly used to delay free radical accumulation and strengthen the oxidative stability [68]. Consumption of natural antioxidants from plant sources is a wise option for the prevention of oxidative stress-related disorders, aswith many chronic, neurodegenerative, and cardiovascular diseases [69]. The antioxidant properties of natural products, like EOs, are related to the nature of the bioactive compounds and sometimes to synergistic effects between them. There are several methodologies widely used to determine antioxidant capacity and, in this work, the free radical scavenging capacity of H. italicum was determined using DPPH assay, RP, ABTS • +, and BCB assay. DPPH radical scavenging assay determines the ability of samples to donate an electron and scavenge DPPH radicals. RP of phytochemicals is associated with antioxidant capacity, since it is related to their ability to transfer electrons. One of the most frequently used organic radicals for the evaluation of antioxidant efficiency of pure substances and complex systems are stable synthetic ABTS • +. The BCB assay determines lipid peroxidation in oil-in-water emulsions; it measures the loss of the yellow color of β-carotene due to its reaction with formed linoleic acid radicals caused by oxidation, but this process could be minimized by the presence of antioxidants. EOs contain antioxidants which act in a hydrophobic environment, inhibiting lipid peroxidation and scavenge lipid peroxyl radicals, thus preventing the propagation of free radical-mediated chain reactions [70].
In the study by Kladar et al. (2015) [46], EO of H. italicum subsp. italicum exhibited IC50 value of 1.37 mg/mL to inhibit DPPH radicals, which was weak in comparison with that ofthe ethanol extract of the same plant. Furthermore, the activity obtained by β-carotene bleaching assay was found to be 96.58 µmoLTE/100 g. Poli et al. (2003) [71] performed investigation where DPPH test was performed on H. italicum dried flower heads from plants grown in different areas of north-east Italy under different growing conditions. The DPPH assay showed that all extracts in this study showed minimal antioxidant activity, even at the lowest test concentration of 5 mg/mL. Also, the β-carotene bleaching test was applied in the same study, with respect to the variation in antioxidant activity after 28 and 56 h, and exhibited that the relative antioxidant activities values for the samples of dried flower heads from wild H. italicum, grown in experimental open fields, and commercial drug were quite similar at the first step (28 h). A comparison with the reference compound (BHA) showed that the antioxidant activity of these Helichrysum extracts was around 19% lower than that of BHA at the same concentration. However, dried flower heads from H. italicum produced in the nursery, exhibited lower activity than the other samples (43% lower than BHA at the same concentration). After 56 h, only the sample grown in experimental open fields maintained relatively constant relative antioxidant activities values, with a 6% drop from the first step (28 h).
α-Glucosidase, an enzyme located in the small intestine tract, plays a role in the final step of carbohydrate digestion, the breakdown of starch and disaccharides to glucose [72]. The enzyme α-glucosidase results in increased blood glucose levels. Inhibition of this enzyme regulates the liberation of D-glucose from complex carbohydrates [73].
Protein denaturation occurs when proteins lose their tertiary and secondary structure after the addition of external stress or compound, such as concentrated inorganic salt, an organic solvent, or heat, and consequently, proteins lose their biological function. Denaturation of protein tissues is one of the most causes of inflammation [74]. Djihane et al. (2017) [43] published a study where diclofenac sodium was found to be less effective when compared with that of H. italicum EO. The anti-inflammatory effect could be caused due to the synergistic effect rather than single constituent in H. italicum essential oil.

Conclusions
The EO of flowering aerial parts of H. italicum introduced from Bosnia and Hercegovina and grown in the Serbia showed an interesting chemical composition. The results from this study showed that the selected five molecular descriptors were adequate in predicting the retention indices of the observed chemical constituents in H. italicum EO. The coefficients of determination for training cycle were 0.964 for compounds found in H. italicum EO. Furthermore, the results suggest that this EO has a potential to be used as flavoring agent, natural antioxidant, and as a value-added bioactive ingredient in food and pharmaceutical products. Considering the demonstrated influence of the chemical composition of H. italicum EO on its biological properties, plant chemotypes may significantly influence the applications of this EO.