Chemical Composition and in Vitro Antimicrobial, Cytotoxic, and Central Nervous System Activities of the Essential Oils of Citrus medica L. cv. ‘Liscia’ and C. medica cv. ‘Rugosa’ Cultivated in Southern Italy

Citrus medica cv. ‘liscia’ and C. medica cv. ‘rugosa’ are two taxa of citron, belonging to the biodiversity of South Italy, in particular of Amalfi Coast, in the Campania region. The chemical composition of the essential oils (EOs) from fruit peels of both C. medica cultivars was studied by gas chromatography (GC) and gas chromatography-mass spectrometry (GC-MS) analyses. In all, 100 compounds were identified, 82 for C. medica cv. ‘liscia’, accounting for 91.4% of the total oil, and 88 for C. medica cv. ‘rugosa’, accounting for 92.0% of the total oil. Monoterpene hydrocarbons are the main constituents in both oils of C. medica cv. ‘liscia’ (79.1%) and C. medica cv. ‘rugosa’ (80.2%). In both oils, limonene (67.2%–62.8%) and camphene (8.5%–10.9%) are the main constituents. The antimicrobial activity of the EOs was assayed against some bacterial strains: Bacillus cereus (DSM 4313), Bacillus cereus (DSM 4384), Staphylococcus aureus (DSM 25693), Pseudomonas aeruginosa (ATCC 50071), and Escherichia coli (DSM 8579). Low concentrations of C. medica cv. ‘rugosa’ EO showed an inhibitory effect on P. aeruginosa and higher concentrations inhibited more B. cereus (4384) and E. coli than S. aureus. The cytotoxicity of the EO was evaluated against SH-SY5Y cell line. The influence of the EO on the expression of adenylate cyclase 1 (ADCY1) was also studied. The antimicrobial activity registered confirm their traditional uses as food preserving agents and led us to hypothesize the possible use of these oils as antimicrobials. The alterations in ADCY1 expression suggested a role for limonene in effects on the central nervous system.


Introduction
Citron, native to Southeast Asia, was imported to the Mediterranean around 300 B.C. Probably, it arrived in Italy through the Hebrews who introduced the cultivation of citron on the Calabrian coasts, Amalfi Coast, and Garda Lake [1,2].
Two local cultivars of Citrus medica L. are grown on the Amalfi Coast: C. medica cv. 'liscia', known by the vernacular name of 'cedro', and C. medica cv. 'rugosa', known as 'ponsino'. These two cultivars contributed to the agricultural biodiversity of this area, as well as other Citrus species. However, their

No.
Compound a Linear retention index on a HP-5MS column; b Linear retention index on a HP Innowax column; c Identification method: 1 = linear retention index; 2 = identification based on the comparison of mass spectra; 3 = Co-Gas chromatography retention time identical to authentic compounds; -: not detected; Tr = trace (<0.1%).

Antibacterial Activity
Using the inhibition halo technique, we evaluated the antimicrobial activity of the EOs of both C. medica cultivars. Among the Gram-positive bacteria, Bacillus cereus 4384 and Staphylococcus aureus were the more sensitive bacterial strains to C. medica cv. 'liscia' EO ( Figure 1).
Low concentrations of C. medica cv. 'rugosa' and cv. 'liscia' EOs showed strong effect on all tester strains (inhibition halos ranging between 8.5 and 10 mm, Figures 1 and 2). The effect was higher also respect to the antibiotics used as a control. B. cereus 4384 resulted in the most sensitive strain; S. aureus was the most sensitive at the lowest amount of Citrus medica cv. 'liscia'. In any case, C. medica cv. 'rugosa' EO showed a greater inhibitory power than C. medica cv. 'liscia' EO on all bacterial strains and already at low concentrations ( Figure 2). Three microliters of the two EOs seemed to give the most effective antimicrobial activity. At this concentration, the behavior exhibited by the pathogen strains was different in terms of sensitivity. Certainly, C. medica cv. 'rugosa' EO was much more effective than the EO obtained from C. medica cv. 'liscia'. P. aeruginosa and E. coli were the most sensitive strains to this last EO, with inhibition halos not inferior to 27 mm. A strong activity was also exhibited by C. medica cv. 'rugosa' EO against S. aureus. C. medica cv. 'liscia' EO was effective, although in not such a marketable mode against the strains, with inhibition halos never superior to 22 mm. The two strains of B. cereus used in the experiments (4313 and 4384) reflected, in some way, the different behavior exhibited by the microorganisms used as tester. Both strains were sensitive to the two EOs. B. cereus 4313 showed the same behavior, giving similar inhibition halo; on the contrary, B. cereus 4384 showed a different behavior, resulting more sensitive to the C. medica cv. 'rugosa' EO respect to the EO obtained from C. medica cv. 'liscia' (∆ = 19%). Minimum inhibitory concentration (MIC) values reported a remarkable action of the two EOs. C. medica cv. 'liscia' EO (MIC = 0.1 µL/mL, Table 2) was more effective than that of C. medica cv. 'rugosa' (MIC ranging between 0.1 and 0.8 µL/mL), except than against B. cereus 4384 (MIC = 0.1 µL/mL) and much more effective against the antibiotic gentamycin. In all cases, the results of the MIC assay indicated that the antimicrobial activity of the two EOs was much more effective than their two main components, limonene and camphene.
'rugosa' EO showed a greater inhibitory power than C. medica cv. 'liscia' EO on all bacterial strains and already at low concentrations ( Figure 2). Three microliters of the two EOs seemed to give the most effective antimicrobial activity. At this concentration, the behavior exhibited by the pathogen strains was different in terms of sensitivity. Certainly, C. medica cv. 'rugosa' EO was much more effective than the EO obtained from C. medica cv. 'liscia'. P. aeruginosa and E. coli were the most sensitive strains to this last EO, with inhibition halos not inferior to 27 mm. A strong activity was also exhibited by C. medica cv. 'rugosa' EO against S. aureus. C. medica cv. 'liscia' EO was effective, although in not such a marketable mode against the strains, with inhibition halos never superior to 22 mm. The two strains of B. cereus used in the experiments (4313 and 4384) reflected, in some way, the different behavior exhibited by the microorganisms used as tester. Both strains were sensitive to the two EOs. B. cereus 4313 showed the same behavior, giving similar inhibition halo; on the contrary, B. cereus 4384 showed a different behavior, resulting more sensitive to the C. medica cv. 'rugosa' EO respect to the EO obtained from C. medica cv. 'liscia' (Δ = 19%). Minimum inhibitory concentration (MIC) values reported a remarkable action of the two EOs. C. medica cv. 'liscia' EO (MIC = 0.1 μL/mL, Table 2) was more effective than that of C. medica cv. 'rugosa' (MIC ranging between 0.1 and 0.8 μL/mL), except than against B. cereus 4384 (MIC = 0.1 μL/mL) and much more effective against the antibiotic gentamycin. In all cases, the results of the MIC assay indicated that the antimicrobial activity of the two EOs was much more effective than their two main components, limonene and camphene.     Limonene, C. medica cv. 'liscia', and C. medica cv. 'rugosa' EOs were evaluated for their ability to inhibit the growth of human neuroblastoma cell line (SH-SY5Y). The EOs and limonene revealed different cytotoxic activities. Limonene and C. medica cv. 'rugosa' EOs showed an IC 50 > 2000 µg/mL, instead C. medica cv. 'rugosa' EO showed an IC 50 of 718.2 µg/mL.

MIC (μL)
Treatment of SH-SY5Y neuroblastoma cells with 800 µg/mL of limonene for 24 h resulted in a low cytotoxic activity. However, treatment with 800 µg/mL of C. medica cv. 'liscia' EOs resulted in a stronger cytotoxicity than C. medica cv. 'rugosa' EO with 38% cell death ( Figure 3).
Treatment of SH-SY5Y neuroblastoma cells with 800 μg/mL of limonene for 24 h resulted in a low cytotoxic activity. However, treatment with 800 μg/mL of C. medica cv. 'liscia' EOs resulted in a stronger cytotoxicity than C. medica cv. 'rugosa' EO with 38% cell death ( Figure 3).

ADCY1: Western Blotting Analysis
We investigated the effects of limonene, C. medica cv. 'liscia' and C. medica cv. 'rugosa' EOs in SH-SY5Y human neuroblastoma cells. Representative Western blots and quantitative densitometric analysis for adenylate cyclase 1 (ADCY1) protein expression in SH-SY5Y following exposure to different concentrations of limonene and EOs are shown in Figure 4. Treatments of SH-SY5Y neuroblastoma cells with 800 and 50 µg/mL of limonene for 24 h significantly influenced ADCY1 expression in different way: high concentrations appear to increase ADCY1 expression, instead low concentrations reduced the ADCY1 expression ( Figure 4A). However, treatments with 400, 200, 100, 50 µg/mL of C. medica cv. 'liscia' and C. medica cv. 'rugosa' EOs appear to influence significantly ADCY1 expression with an over expression and a down expression of ADCY1, respectively ( Figure 4B,C).

Discussion
There were no relevant quali-quantitative differences in the composition of the two EOs, neither with regard to the percentages of the classes of compounds nor to the main components. In both oils, limonene and camphene are the main components, even if the composition of the EO of C. medica cv. 'rugosa' is more complex, but many of its constituents are present in very low percentages or even in traces. Values are the mean ± SD in each group (n = 3). * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, compared to control (ANOVA followed by Dunnett's multiple comparison test).

Discussion
There were no relevant quali-quantitative differences in the composition of the two EOs, neither with regard to the percentages of the classes of compounds nor to the main components. In both oils, limonene and camphene are the main components, even if the composition of the EO of C. medica cv. 'rugosa' is more complex, but many of its constituents are present in very low percentages or even in traces.
The comparison with the available literature concerning the chemical composition of the EO of citron from other countries, showed substantial differences. Limonene percentages found (67.2% and 62.8%) are significantly higher than the values recorded in different parts of the world or even in Italy: South Korea, 52.44% [11]; China, 33.84% [12]; Santa Maria del Cedro (Italy), 59% [1]; and Iran, 56.6% [13]. Other notable differences relate to the content in γ-terpinene exceeding 20% and of αand β-pinene, reported between 7.73% [12] and 16.3% [13]. Camphene range is 8.5%-10.9% (cv. 'liscia' and cv. 'rugosa', respectively) in samples studies in this research, but this compound is practically absent or in traces in the oils reported in literature.
The antimicrobial activity of the two EOs was evaluated against different microorganisms: Bacillus cereus (DSM 4313 and DSM 4384), an agent of different foodborne illnesses which can be found in raw and processed foods; Staphylococcus aureus which plays a significant role in invasive skin disease, and actually representing one of the most significant emerging pathogens; an enterotoxigenic strain of Escherichia coli, generally involved in episodes of food poisoning; and Pseudomonas aeruginosa, concerned to different urinary, pulmonary, skin, ear and eye infections.
The modest inhibitory activity exhibited by the two EOs against the Gram-negative strains could be due to the different structures of the cell walls of Gram-positive and Gram-negative bacteria [14]. Our results supported previous studies [15], which reported a different sensitivity shown by the two strains of B. cereus against EO of Hypericum perfoliatum.
Some authors reported the antimicrobial activity of C. medica against different pathogens. Sah and coworkers [16] evaluated the antimicrobial activity of fruit juice and ethanol extract of different plant portion of Citrus medica which resulted to be less effective than the EOs of Citrus medica cv. liscia and rugosa evaluated in our experiments. Belletti and coworkers evaluated the effects of citron EO on microbial spoilage and growth and survival of pathogenic microorganisms during storage of a fruit based salad, reporting the capability of these substances to prolong the microbial shelf life of these products [17]. Therefore, the values of MIC observed testing limonene and camphene, the two most abundant components present in the two EOs, seemed to corroborate the hypothesis-almost for all pathogens tested-that the activity is due to a synergy among different components, more than to a specific component. In fact, MIC of limonene was always 1 µL, ten-fold higher than that of the two EOs; camphene, generally recognized as good antimicrobial agent [18], was less effective (MIC > 6 µL) except than against E. coli (MIC = 0.084 µL). Considering such value and comparing it with the MIC of the two EOs of Citrus medica, we could hypothesize a strong influence of camphene on the antimicrobial. On the other hand, these two monoterpenes generally do not exhibit a strong antimicrobial activity [19] per se. Limonene was indicated for its inhibitory activity against the Gram-negative pathogens, P. aeruginosa and E. coli, but at higher concentrations [20][21][22]. Limonene exists in two enantiomeric forms, with different biological activity. Aggarwal and coworkers have demonstrated that (S)-(−)-limonene had less antimicrobial activity than (R)-(+)-limonene. In fact, the first failed to inhibit many bacterial strains and the majority of the fungal strains, except Microsporum gypseum. In contrast, (R)-(+)-limonene was highly active against a wide spectrum of microorganisms [23].
The antimicrobial activity of the two EOs might be due to a global synergy among different singular components, more than to the presence of a specific component.
It is remarkable the stronger antimicrobial activity exhibited by the two EOs. This is still more noteworthy, considering their MIC values. The increasing resistance of microorganisms to conventional chemicals and drugs has prompted scientists to search for novel sources of biocides with broad-spectrum activities. In this concern, the activity of the two EOs studied comforts us towards an ever-greater increase of studies and research aimed at the discovery, exploitation, and potential use of natural substances capable of effectively countering the action of pathogenic microorganisms.
The cytotoxic activity of the two EOs and limonene were evaluated in human neuroblastoma cell line (SH-SY5Y). The IC 50 values were >700 µg/mL, indicating that the substances were not cytotoxic as judged by the criterion set by the National Cancer Institute that stated that extracts with IC 50 < 20 µg/mL were considered to be cytotoxic against the treated cells [24].
However, comparing the IC 50 values, our findings indicated that C. medica cv. 'liscia' EO is more cytotoxic than C. medica cv. 'rugosa' EO and limonene. The cytotoxicity can probably be attributed to a synergistic activity of limonene and other minor components. Monajemi and coworkers [13] reported that the EO of C. limon with large amounts of limonene (98.4%) was less cytotoxic than C. medica with low content of limonene (56.6%) on MCF-7 and HeLa cell lines. Our results agree with previous studies reporting that low concentrations of limonene were ineffective in cell death in SH-SY5Y cells [7,25].
Many plant species with EOs are used as sedatives, hypnotics, and tranquilizers and they may be useful in treating disorders of the central nervous system [26]. Different studies reported a well-established role for adenylyl cyclase in the regulation of synaptic plasticity, memory, and other multiple brain processes. Moreover, cross-talk between the cyclic adenosine-3 ,5 -monophosphate (cAMP) signal transduction system and other signaling pathways, such as the extracellular signal-regulated kinase/mitogen-activated protein (ERK/MAP) regulatory system, has been described [27,28]. In this perspective, we evaluated the possible influence of limonene and C. medica cv. 'liscia' and C. medica cv. 'rugosa' EOs on ADCY1 expression in SH-SY5Y cells, a human-derived neuroblastoma cell line often used as an in vitro neuronal model [29]. Adenylyl cyclase is an enzyme that regulates the physiological effects of numerous drugs and hormones through the production of cAMP. Clinical and epidemiologic research provided suggestive evidence regarding the association between adenylyl cyclase activity and major depression [30,31]. Moreover, platelet adenylyl cyclase activity might serve as a biological marker for major depression and the therapeutic effect of antidepressants [31][32][33][34].
Our results showed that the higher concentration of limonene (800 µg/mL) appears to increase ADCY1 expression in SH-SY5Y cell and, consequently, the intracellular production of cAMP. Our results agree with results of Watt and coworkers [35] that showed that some volatile substances stimulate cAMP-response element (CRE)-mediated transcription in olfactory sensory neurons through Ca 2+ activation of the ERK/MAPK/p90rsk (Ribosomal Protein S6 Kinases, 90-kDa) pathway.
C. medica cv. 'liscia' EO showed similar effects with all concentrations tested, probably because its main component is limonene (67.2%). Limonene is reported to increase cytosolic cAMP concentration and induce activation of protein kinase A and phosphorylation of cAMP-response element-binding protein in Chinese hamster ovary cells transfected with the human adenosine A2A receptor gene [36]. On the contrary, treatment with C. medica cv. 'rugosa' EO showed an inhibition on ADCY1 expression in SH-SY5Y cell, possibly due to the complex phytochemical composition of this EO. These results showed that the use of both EOs could increase intracellular production of cAMP and have an excitatory effect on the Central Nervous System [26].

Isolation of the Essential Oils
One hundred grams of fruit peels of each cultivar were ground in a Waring blender and then subjected to hydrodistillation for 3 h according to the standard procedure described in the European Pharmacopoeia [37]. The oils were solubilized in n-hexane, filtered over anhydrous sodium sulphate and stored under N 2 at 4 • C in the dark, until tested and analyzed.

GC/MS Analysis
Analysis was performed on an Agilent 6850 Ser. II apparatus (Agilent), fitted with a fused silica DB-5 capillary column (30 m × 0.25 mm i.d., 0.33 µm film thickness, Agilent), coupled to an Agilent Mass Selective Detector MSD 5973; ionization energy voltage 70 eV; electron multiplier voltage energy 2000 V. Mass spectra (MS) were scanned in the range 40-500 amu, scan time 5 scans/s. Gas chromatographic conditions were as reported in the previous paragraph; transfer line temperature, 295 • C.

Identification of the Essential Oil Components
Most constituents were identified by GC by comparison of their Kovats retention indices (Ri) (determined relative to the retention times (tR) of n-alkanes (C10-C35)), with either those of the literature [38][39][40][41] and mass spectra on both columns or those of authentic compounds available in our laboratories by means of NIST 02 and Wiley 275 libraries [42]. The components' relative concentrations were obtained by peak area normalization. No response factors were calculated.

Antimicrobial Activity
To study antimicrobial activity a filter paper disc method was used [43]. The bacteria used in this study included Bacillus cereus DSM 4313, Bacillus cereus 4384, Staphylococcus aureus DSM 25693, Pseudomonas aeruginosa ATCC 50071, and Escherichia coli DSM 8579. Bacteria were purchased from the Deutsche SammLung von Mikroorganismen und Zellkulturen GmbH (DSMZ, Deutsche Sammlung von Mikroorganismen und Zellkulturen, Braunschweig, Germany).
A 100 mg sample of each EO was re-suspended in dimethyl sulfoxide (DMSO) to produce the same concentration of 10 mg/mL, then diluted to be subjected to biological analyses. The strains were incubated in nutrient broth (Oxoid, Milano, Italy) at 37 • C for 18 h. The optical densities of all cultures were adjusted to match a 0.5 McFarland standard of 1 × 10 8 colony-forming units (cfu)/mL. Sterile filter paper discs (5 mm) were impregnated with 1 and 10 µL of the EOs and placed in Petri dishes. Plates were left for 30 min at room temperature under sterile conditions and then incubated at 37 • C for 24 h, and the inhibition halo around the disc was measured. A disc treated with DMSO alone served as the negative control. Tetracycline (7 µg/disc; Sigma) and gentamycin (8 µg/disc; Sigma) were used as reference drugs. The experiments were performed in triplicate and averaged.

Minimum Inhibitory Concentration
For the MIC of the EOs and their main components, limonene and camphene, a modified version of the resazurin microtiter-plate assay of Sarker and coworkers was used [44]. Briefly, different volumes of the two EOs, prepared as above described, and of the pure compounds limonene (dissolved in DMSO, 1:10 v/v), and camphene (dissolved in a mix hexane/DMSO 1:10 v/v), were used and pipetted into the first row of the 96 well plates. To all wells, different volumes of Muller-Hinton broth was added. Two-fold serial dilutions were performed such that each well had 50 µL of the test material in serially descending concentrations. 30 µL of 3.3× strength isosensitised broth and 10 µL of resazurin indicator solution (previously prepared by dissolving 270 mg tablet in 40 mL of sterile distilled water) were added in each well, to reach a final volume/well of 240 µL. Finally, 10 µL of bacterial suspension was added to each well to achieve a concentration of approx. 5 × 10 5 cfu/mL. Each plate was wrapped loosely with cling film to ensure that bacteria did not become dehydrated. In each plate a set of controls were added: a column with a ciprofloxacin (1 mg/mL in DMSO, Sigma) as positive control; a column with all solutions with the exception of the test compound; a column with all solutions with the exception of the bacterial solution adding 10 µL of nutrient broth instead and a column with DMSO solution as a negative control. The plates were prepared in triplicate, and incubated at 37 • C for 24 h. The growth was indicated by color changes (visually assessed) from blue/dark purple to pink (or colorless). The lowest concentration at which color change occurred indicated the MIC value.

MTT Bioassay
Human neuroblastoma cancer cells (SH-SY5Y) were plated (5 × 10 3 ) in 96-well culture plates in 150 µL of culture medium and incubated at 37 • C in humidified 5% CO 2 . The day after, a 150 µL aliquot of serial dilutions of limonene and EOs (800-50 µg/mL) was added to the cells and incubated for 24 h. DMSO alone was used as control. Cell viability was assessed through 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay. Briefly, 30 µL of MTT (5 mg/mL) was added and the cells incubated for additional 3 h. Thereafter, cells were lysed and the dark blue crystals solubilized with 30 µL of a solution containing 50% v/v N,N-dimethylformamide and 20% w/v SDS with an adjusted pH of 4.5. The optical density (OD) of each well was measured with a microplate spectrophotometer (Thermo Scientific Multiskan GO, Monza, Italy) equipped with a 520 nm filter. Cell viability in response to treatment was calculated as a percentage of control cells treated with DMSO at the final concentration (0.1%): viable cells = (100 × OD treated cells)/OD control cells [45].

Extraction Proteins and Western Blotting
Cells were treated with different concentrations (800-100 µg/mL) of limonene and EOs. After 24 h, cells were collected and lysed using LaemmLi buffer to extract total proteins. For Western blot analysis, an aliquot of total protein was run on 8% SDS-PAGE gel and transferred to nitrocellulose. Nitrocellulose blots were blocked with 10% nonfat dry milk in Tris buffer saline 0.1% Tween-20 over night at 4 • C and incubated with primary anti-ADCY1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 3 h at room temperature. Immunoreactivity was detected by sequential incubation with horseradish peroxidase-conjugated secondary antibody (Amersham Biosciences, Pittsburgh, PA, USA) and enhanced chemiluminescence reagents (ImmunoCruz, Santa Cruz Biotechnology) [46].

Statistical Analysis
All experiments were carried out in triplicate. Data of each experiment were statistically analyzed using GraphPad Prism 6.0 software (GraphPad Software Inc., San Diego, CA, USA) followed by comparison of means (two-way ANOVA) using Dunnett's multiple comparisons test, at the significance level of p < 0.05.

Conclusions
We characterized the chemical composition of the EOs of two C. medica cultivars, characteristics of Amalfi Coast. These oils present a new composition, with high presence of limonene, which seems a 'fingerprint' of these oils. The EOs showed higher antimicrobial activity at the tested doses. The EOs showed higher antimicrobial activity than antibiotics used as control. The presence of different components present in the two oils, more than the high content of limonene and camphene might be responsible for this EOs activity. This finding led us to hypothesize the possible use of these oils as antimicrobials and confirm their traditional uses as food preserving agents.
The alterations in ADCY1 expression suggested a major role for limonene in effects on the central nervous system. The precise mechanism of the biochemical effects of Citrus medica EOs is unclear; however, ADCY1 pathway might be involved. Moreover, the chemical and biological study of these EOs contribute to the knowledge and promotion of two cultivars belonging to the biodiversity of South Italy.

Conflicts of Interest:
The authors declare no conflict of interest.

Abbreviations
The following abbreviations were used in this manuscript: