Composition and antimicrobial activity of Pastinaca sativa subsp. sativa, P. sativa subsp. urens and P. hirsuta essential oils

ABSTRACT Composition and antimicrobial activity of root, leaf, stem, flower and fruit essential oils from cultivated Pastinaca sativa subsp. sativa, and its two wild-growing relatives P. sativa subsp. urens and P. hirsuta (Apiaceae) were investigated. Twenty-nine hydrodistilled essential oils of plants from different localities and/or years were analysed by GC-FID and GC/MS. Dominant in root oils was myristicin (P. sativa) or apiole (P. hirsuta), in leaf and stem oils myristicin (cultivated plants) or γ-palmitolactone (wild-growing plants) and in flower and fruit oils aliphatic esters. Multivariate statistics (PCA, nMDS, UPGMA clustering) generally revealed separation of oils of investigated Pastinaca taxa and demonstrated their chemosystematic significance. One oil per each organ of all three plants (fifteen in total) was tested using microdilution method for activity against Candida tropicalis, C. parapsilosis, C. krusei, C. glabrata, C. albicans, Staphylococcus aureus, Bacillus cereus, Listeria monocytogenes, Escherichia coli, Salmonella Typhimurium and Enterobacter cloacae; MIC = 0.25–8 mg/mL, MBC(MFC) = 0.5–16 mg/mL.


Introduction
It is widely recognized that consumption of vegetables and spices from the Apiaceae family has numerous health benefits, which can be least partly attributed to the presence of essential oils. Therefore, isolated essential oils of these plants can be applied for various medicinal purposes, and also can be added to foods, for example, to prevent the growth of the spoilage microorganisms (1,2).
The constantly growing demand for food has critically increased the production of agricultural waste and consequently the costs of its disposal (3). In agreement with circular economy perspective, aromatic waste biomasses (such as those resulting from the growing of aromatic food crops) can be used for the obtaining of bioactive essential oils (2). Wildgrowing relatives of cultivated plants are also being investigated as possible new sources of such bioactive products.
Pastinaca sativa subsp. sativa L., Apiaceae (parsnip) is a biennial plant, which is cultivated mainly in the temperate regions of the world. Demand for this crop continuously increases year by year (4,5). Parsnip root is a well-known vegetable, which can be consumed raw, cooked, baked, fried or roasted. Most commonly, it is used as a constituent of various soups (both dried as a seasoning and fresh), but can be also added to stews, salads, casseroles, pies, puddings, etc. (5). The root of the best quality is obtained from the plants from the first year, in which usually only leaf rosette is formed (6). These parsnip leaves can be also added to soups (5), but in most cases, they are considered for agricultural waste. Furthermore, in order to obtain fruits for reproduction, parsnip is grown for 2 years and in that case, all other plant organs are also agricultural waste. Different parts of parsnip are also used in folk medicine. For example in Serbia, infusion prepared from roots, leaves or fruits is taken for its digestive and diuretic properties and to improve appetite and milk production (7). Similarly in Italy, infusion from roots or leaves is used as dietetic, cholagogue and diuretic (8). Previously, the essential oils of different parsnip organs, originating from, e.g. Germany, the USA, Great Britain and Denmark were investigated (9,10). Furthermore, it was shown that the essential oil of roots of this plant possesses anticandidal activity (11).

Obtaining of essential oils
The air-dried material was cut (leaves and flowers) or ground in an electric grinder (roots, stems and fruits) and hydrodistilled using Clevenger-type apparatus for 2.5 h. Obtained essential oils were dried over anhydrous sodium sulphate and kept at 4°C, in dark, in sealed glass vials until analysis. The oil yields are included in Table 1.   (18), as well as by the comparison of their RIs with those from the literature (18) and those deposited at the NIST Chemistry WebBook website. Relative percentages of the compounds were calculated based on the peak areas from the FID data.

Multivariate statistical analysis
Chemical composition of the essential oils was analysed using multivariate statistical methods: principal component analysis (PCA), non-metric multidimensional scaling (nMDS) and unweighted pair-group arithmetic averages (UPGMA) clustering. The analyses included the compounds that were present in at least one essential oil in the quantity ≥1%. In order to reduce the large differences between these amounts, the data were transformed using log and arcsin transformations, as well as coding before the analyses (19,20). The best results (presented in further text) were obtained using coding. The codes were assigned in the following way: value 1 for 0% (i.e. when the constituents were not detected), value 2 for traces (<0.1%), value 3 for quantities ≥0.1% and <1%, value 4 for quantities ≥1% and <5%, value 5 for quantities ≥5% and <10%, value 6 for quantities ≥10% and <20%, value 7 for quantities ≥20% and <40%, value 8 for quantities ≥40% and <60%, value 9 for quantities ≥60% and <80%, and value 10 for quantities ≥80% (20). PCA was used to find the variables that contribute the most to the position of the investigated samples in the PCA ordination graph (based on the factor coordinates, i.e. loading plot comparison). nMDS was performed to graphically delineate dissimilarities and grouping among the investigated samples, whereas UPGMA was used for the agglomerative hierarchical cluster analysis. nMDS and UPGMA were based on the Bray-Curtis pairwise distance matrix (19,20). The analysis was performed using Statistica 6.0 (StatSoft Inc., Tulsa, OK, USA).

Microbial cultures
The anticandidal activity of the essential oils was tested against three standard strains: C.

Anticandidal activity
Minimum inhibitory concentrations (MICs) and minimum fungicidal concentrations (MFCs) of the essential oils were determined by the modified microdilution technique in 96-well microtiter plates (Spectar, Čačak, Serbia) as described previously (21). Briefly, yeast cultures were diluted using sterile saline to a concentration of approximately 1.0 × 10 5 CFU/ well. Then, microplates containing fungal cells were incubated with serial dilutions of the oils at 37°C for 24 h. MICs represented the lowest concentrations of the oils at which no microscopic growth was observed. MFCs were determined after serial subcultivations of 10 µL into microtiter plates containing 100 µL of broth/well and further incubation at 37°C for 24 h. MFCs represented the lowest concentrations with no visible growth, indicating the killing of 99.5% of the original inoculum. Ketoconazole was used as a positive control.

Antibacterial activity
MICs and minimum bactericidal concentrations (MBCs) of the essential oils were determined by microdilution method using 96-well microtiter plates (22). Briefly, the bacterial suspensions were adjusted with sterile saline to a concentration of 1.0 × 10 5 CFU/mL and added to serially diluted essential oils. The lowest concentrations without visible growth (under a binocular microscope) after 24 h incubation at 37°C were defined as MICs. The MBCs were determined by serial sub-cultivations of 10 μL into microtiter plates containing 100 μL of broth per well and further incubation for 24 h. The lowest concentrations with no visible growth were defined as the MBC, indicating the killing of 99.5% of the original inoculum. Streptomycin and ampicillin were used as positive controls.

Chemical composition of Pastinaca essential oils
Twenty-nine Pastinaca essential oils were investigated (Table 1). These include eleven P. sativa subsp. sativa oils (plants from four localities; the oils of roots and leaves were obtained from plants from both the first and the second year, and of other organs from plants from the second year), ten P. sativa subsp. urens oils (plants from two localities) and eight P. hirsuta oils (plants from one locality, collected in two different years). The yields of analysed essential oils are given in Table 1. GC-FID and GC/MS analysis revealed the presence of twentytwo (sat5; acronyms are explained in Table 1) to fortyfive (sat4) components in eight investigated root essential oils, thirty-four (sat6) to forty-nine (ure2) in five leaf oils, twenty-seven (hir2) to fifty-two (ure2) in six stem oils, thirty (hir2) to forty-five (sat6) in four flower oils and thirteen (hir1) to forty-three (sat7) in six fruit oils. Representative GC-FID chromatograms, which contain numbered peaks of almost all (73 out of the total of 77) identified compounds (except of four long chain alkanes with RIs higher than 2294) are given in Supplemental material ( Figures S1-S8). In one case, identified components accounted for 89.4% of the oil (ure1 stem oil), whereas in all other cases that number was above 90.0%, reaching 98.6% (sat5 root oil).

The composition of the root essential oils
Chemical composition of investigated Pastinaca root essential oils is presented in Table 2. Phenylpropanoid myristicin dominated in the root oils (39.7-82.5%) of both investigated P. sativa subspecies (i.e. in the oils of cultivated sat1, sat3-5 and wild-growing ure1 and ure2 samples). The second most abundant component in the root oils of all P. sativa subsp. sativa samples from the first year (sat1, sat3 and sat5) and one P. sativa subsp. urens (ure1) sample was monoterpene terpinolene (14.8-28.7%). Interestingly, in the root oil of P. sativa subsp. sativa sample from the second year (sat4), the content of terpinolene significantly decreased (1.2%). In the case of the root oils of wild-growing P. hirsuta (hir1 and hir2 samples), another phenylpropanoid apiole (30.9% and 25.8%) was the dominant and the contents of myristicin were somewhat lower (11.6% and 20.3%). In P. sativa subsp. urens and P. hirsuta root oils notable amounts of polyacetylene (Z)-falcarinol (10.5-25.9%) and γ-palmitolactone (7.9-15.6%; except in ure1 oil) were also determined.
In PCA and nMDS analyses of these results (Figures 1 and S9), the separation of the oils of all three investigated taxa was observed. In PCA, apiole (factor loading −0.91) and octadecanoic acid isomer (−0.89) (present in notable quantities only in P. hirsuta oils) contributed the most to the first principal axis (which explained 52.30% of variation) and αacoradiene (−0.88) (present in amounts other than traces only in P. sativa subsp. urens oils) to the second principal axis (which explained 20.83% of variation). In UPGMA analysis (Figure 1), two separate clusters were formed by P. sativa and P. hirusta samples, however within the P. sativa cluster, investigated P. sativa subsp. sativa sample from the second year was closer to P. sativa subsp. urens samples, suggesting the impact of phenological stage on the essential oils composition (i.e. both P. sativa subsp. sativa sample from the second year and P. sativa subsp. urens samples were collected from the plants in the flowering and fruiting phases).
The oils of the roots from the first year of cultivated P. sativa subsp. sativa were also investigated previously (5,9). For example, in twenty-four root oils (twelve from Germany, seven from the USA, four from Great Britain and one from Denmark), myristicin and terpinolene were also the dominant constituents (together they comprised more than 80%); in six oils, the dominant was myristicin (43.6-66.2%), like in the oils analysed in the current work, while in remaining eighteen oils, terpinolene prevailed (44.5-67.3%) (9).

The composition of the leaf and stem essential oils
The results of GC-FID and GC/MS analysis of investigated Pastinaca leaf and stem essential oils are given in Table 3. Myristicin was the dominant constituent in the leaf oils of cultivated P. sativa subsp. sativa (42.8% and 41.4% in sat5 and sat6 oils) and γ-palmitolactone in the leaf oils of investigated wild-growing Pastinaca taxa 1095 tr  tr  tr  tr  1485  1484  Germacrene D  tr  tr  tr  -----1487 1487 .3% in sat6 and sat7 oils) and γ-palmitolactone (50.6-60.4% in ure1, ure2, hir1 and hir2 oils and also 11.8% and 18.4% in sat6 and sat7 oils) were higher. PCA, nMDS and UPGMA analyses of the leaf essential oils (Figures 2a and S10) resulted in clear separation of the samples of three investigated taxa; however, in nMDS and UPGMA analyses, P. sativa subsp. urens samples were closer to P. hirsuta sample than to P. sativa subsp. sativa samples. The first principal axis in PCA (which explained 42.76% of variation) was mainly defined by γ-palmitolactone (factor loading −0.98) (detected in lower amounts in P. sativa subsp. In PCA, nMDS and UPGMA analyses of the stem oils (Figures 2b and S11), the separation of P. hirsuta and P. sativa samples, as well as of the samples of the two P. sativa subspecies was observed. In PCA, lavadulyl acetate (factor loading 0.98), (Z)-β-ocimene (0.97), terpinolene (0.94) and 2-phenyl ethyl butanoate (0.93) (detected only in P. sativa subsp. urens oils or present in higher amounts in these oils) mostly contributed to the first principal axis (which explained 53.02% of variation) and α-trans-bergamotene (0.98), myristicin (0.97) (present in higher amounts in P. sativa subsp. sativa oils), hexadecanoic acid (−0.92) and γpalmitolactone (−0.92) (not detected or present in lower amounts in P. sativa subsp. sativa oils) to the second principal axis (which explained 34.34% of variation).
Previously, in the leaf and stem oils of cultivated P. sativa subsp. sativa (from Germany), similar relations between the contents of (E)-β-farnesene (17.2% and 8.0%), myristicin (40.7% and 44.0%) and γpalmitolactone (9.8% and 17.9%) were observed. This research also included five leaf and five stem oils of wild-growing P. sativa subsp. sativa, in which the contents of these three compounds were very variable and did not follow the aforementioned rule; the amounts of myristicin varied from 0.6% to 64.2%, of γ-palmitolactone from 5.0% to 24.5% and of (E)-β-farnesene from 1.2% to 24.5%. In addition, these leaf and stem oils contained notable amounts of monoterpenes (Z)-and (E)-β-ocimene and γterpinene (up to 42.2% in the leaf oils and to 55.4% in the stem oils) (10).  Table 1; in PCA, P. sativa samples are marked with empty circles, P. hirsuta samples with coloured circles and compounds (numbers are given in Table 2) with empty squares.   1%). h Numbers in brackets correspond to numbers used to mark the compounds in PCA of the leaf (letter L is added to number of compound) and stem (letter S is added to number of compound) oils (Figure 2; PCA included only those compounds that were present in at least one oil in the quantity ≥1%).  Table 1; in PCA, P. sativa samples are marked with empty circles, P. hirsuta samples with coloured circles and compounds (numbers are given in Table 3) with empty squares.

The composition of the flower and fruit essential oils
Investigated Pastinaca flower and fruit essential oils were dominated by aliphatic esters (  In PCA, nMDS and UPGMA analyses of the fruit oils (Figures 3b and S13), the separation of P. sativa and P. hirsuta samples, but not of the samples of the two investigated P. sativa subspecies was observed. n-Octanol (factor loading −0.98), octyl butanoate (−0.98), myristicin (−0.95) (not detected or present in lower amounts in P. hirsuta oils), hexyl 2-methyl butanoate (0.98), hexyl isovalerate (0.98) and hexyl hexanoate (0.97) (not detected or present in lower amounts in P. sativa oils) contributed the most to the first principal axis in PCA (which explained 61.01% of variation; second principal axis explained only 21.22% of variation).
Previously, in the fruit oil of cultivated P. sativa subsp. sativa (from Germany), like in the current study, octyl butanoate was the most abundant (53.7%). In the oils of five wild-growing P. sativa subsp. sativa fruit samples, notable amounts of both octyl butanoate (28.9-67.2%) and octyl acetate (22.1-53.2%) were detected (10), similarly to P. sativa subsp. urens (ure1) fruit oil analysed in our research. On the other hand, previously investigated P. sativa subsp. urens fruit oil (from Turkey) was dominated by octyl butanoate (79.5%). It contained small amounts of octyl acetate (0.3%) and octyl hexanoate (5.3%) (16), in contrast to ure1 and ure2 fruit oils investigated in our work. The composition of flower and fruit, as well as of root and stem oils of P. hirsuta is in agreement with previous findings obtained by Jovanović et al. (17).

Antimicrobial activity of investigated Pastinaca essential oils
In performed statistical analysis, differences in the composition of the essential oils obtained from all three investigated taxa, as well as stability of the composition of the oils from the same taxon (regardless of locality and date of collection), were observed. Therefore, for the investigation of antimicrobial activity, one essential oil per each organ of all three plants was selected.
All fifteen selected essential oils were able to reduce the growth of different Candida strains (Table 5; MIC range 0.25-4 mg/mL; MFC range 0.5-8 mg/mL). The best activity against all six tested strains (C. krusei, C. glabrata, C. tropicalis, C. parapsilosis and two strains of C. albicans) was observed for the root oils of P. sativa subsp. sativa (sat5) and P. sativa subsp. urens (ure1) (MIC range 0.25-1 mg/mL; MFC range 0.5-2 mg/mL), as well as for the leaf and flower oils of all three taxa (sat6, ure1 and hir2) and the stem oil of P. sativa subsp. urens (ure1) (MIC range 0.5-1 mg/mL; MFC range 1-2 mg/mL); remaining six oils showed such activity on two to five tested strains. Among investigated Candida strains, C. parapsilosis (standard strain) was the most sensitive to examined essential oils (MIC range 0.25-1 mg/mL; MFC range 0.5-2 mg/mL), followed by C. krusei (clinical isolate) and C. albicans (standard strain and clinical isolate) (MIC range 0.5-2 mg/mL; MFC range 1-4 mg/mL). Previously, for the root oils of P. sativa subsp. sativa somewhat lower activity against standard strains and clinical isolates of these same Candida species was demonstrated (MIC range 0.625-2.5 mg/mL; MFC range 1.25-5 mg/mL) (11).
Candida infections are usually localized at mucous membranes, including upper parts of the gastrointestinal tract. In addition, E. coli, L. monocytogenes and S. typhimurium are common sources of intestinal infections (29). Thus, the consumption of root and leaves of cultivated parsnip could contribute to normal functioning of the gastrointestinal tract, both due to the effect of their essential oils against these pathogens (demonstrated in this work) and their ability to act as digestive and cholagogue (based on the use in folk medicine) (7,8).
Furthermore, E. coli is a common pathogen of the urinary tract and Candida spp. can also cause urinary infections (29). It should be noted that root and leaves of cultivated parsnip are known diuretics and can be used for the flushing of the urinary tract (7,8). Besides direct Table 5. Anticandidal activity of the essential oils of three investigated Pastinaca taxa and ketoconazole (mg/mL).  effect on the pathogens (such as E. coli and Candida spp.), increased urine flow is also important for the treatment of these infections. Therefore, our findings could at least partly explain the traditional use of cultivated parsnip, P. sativa subsp. sativa root and leaves, as well as fruits, in the treatment of some gastrointestinal and urinary tract diseases.
Obtained results provide a good basis for further investigation of the essential oils of all three investigated Pastinaca taxa, with the aim of their medicinal application due to demonstrated antimicrobial properties, as well as for prevention of the food contamination with some foodborne pathogens, particularly B. cereus and E. coli (29).

Conclusions
In this work, Pastinaca sativa subsp. urens root, leaf, stem and flower essential oils and P. hirsuta leaf oil were investigated for the first time, whereas the knowledge about other tested essential oils of wild-growing Pastinaca taxa, as well as of cultivated parsnip, P. sativa subsp. sativa was significantly complemented. Multivariate statistical analysis of the essential oils composition revealed that in the most cases all three investigated Pastinaca taxa were separated and that the samples of the P. sativa subspecies were located near each other. These findings together with the fact that observed positions of the samples were not significantly influenced by locality and date of collection, suggest chemosystematic significance of the essential oils composition for the three investigated taxa. Also, it can be concluded that cultivated parsnip, as well as its two wild relatives represent sources of statistically, chemically different essential oils (despite a few evident similarities in the composition of some dominant compounds) with interesting antimicrobial activity. It should be particularly emphasized that this work provides the basis for exploitation of waste biomass resulting from parsnip cultivation, which is in agreement with resources recovery and agricultural waste valorisation principles.