Uncovering the antifungal components from turmeric (Curcuma longa L.) essential oil as Aspergillus flavus fumigants by partial least squares

Abstract

Highly efficient and eco-friendly antifungal fumigants are desirable in food and crop production. Although turmeric (Curcuma longa L.) essential oil is known to have a potent antifungal effect, the quality of whole essential oil can be unstable, leading to unreliable antifungal activities. The aim of the present study is to uncover the active individual compounds in turmeric essential oil that provide the antifungal properties using a convenient chemometric model. The Aspergillus flavus inhibition activities of essential oils derived from 24 batches of Curcuma longa L. were evaluated, with various fumigation concentrations from 50 μL to 500 μL essential oils per plate. Meanwhile, eighteen volatile compounds were identified by static headspace gas chromatography-mass spectrometry. To combine the antifungal activities and chemical profiles with the spectrum-effect relationship based on the partial least squares model, three volatile compounds (i.e., eucalyptol, beta-pinene and camphor) were identified and verified as the most potent antifungal compounds by their higher contribution factor to antifungal indexes. Thus, this research will provide a useful approach to screen bioactive compounds, and these three compounds are promising antifungal alternatives to conventional treatments for the control of A. flavus contamination.

---

1 Introduction

Contamination of food, feedstuffs, crops and medicinal plants with pathogenic fungi is a global public health issue, and reduces food and crop production by 20% in developing countries.^1–4 The aflatoxigenic species of Aspergillus section Flavi, such as A. flavus, are particularly problematic and can cause severe damage to crops both pre- and post-harvest. These fungi produce highly toxic secondary metabolites called aflatoxins, which are classified as group 1 human carcinogens by the International Agency for Research on Cancer.^5–7 Aflatoxins have been implicated in many serious diseases, including hepatocellular carcinoma, acute hepatitis, Reye’s syndrome, cirrhosis in malnourished children, and kwashiorkor.^8,9 Among the aflatoxins, AFB₁ is a particularly toxic and potent hepatocarcinogenic.¹⁰ Due to the high health risk caused by A. flavus contamination, there is much interest in approaches to reduce or eliminate toxigenic fungi and mycotoxins.

There are a number of problems with the current strategies to eliminate or inactivate aflatoxigenic fungi. Although some studies have successfully reduced aflatoxin contamination by co-cultivation with atoxigenic strains, the potential side effects of the introduction of atoxigenic fungi are unclear. Furthermore this approach is not suitable for preventing contamination during the post-harvest period.¹¹ In view of the inefficiency for the removal of aflatoxigenic fungi of physical approaches involving heat, UV light or ionizing radiation treatment,¹² chemical treatments became indispensable for the post-harvest contamination. However, these current hypertoxic chemical agents such as hydrolytic agents or pesticides may leave a high concentration of residues on foodstuffs.¹³ Moreover, many fungicides are now prohibited from use on several crops, thus alternative treatments for the control of mycotoxigenic fungi are required.⁷ Accordingly, in recent years, the development of plant products with antimicrobial properties has been targeted as a viable strategy for food preservation against mycotoxin. Essential oils derived from natural products are regarded as promising natural antifungal agents because of their confirmed antifungal properties, high yields, aromas, and flavors.^14–16 More importantly, essential oils with highly volatile compounds are liable to disperse fast, leaving bare residues in the treated commodities. Many essential oils are widely regarded as safe agents, and possess lower toxicity than commercial pesticides,¹⁷ numerous studies have documented the antifungal effects of essential oils from cinnamon, clove, eucalyptus, Zingiber officinale and Litsea cubeba both in vitro and in vivo.^18–20

Turmeric, which is sourced from the roots of Curcuma longa L. (family Zingiberaceae), is widely used in Southeast Asia for therapeutic use and in food.²¹ Turmeric essential oil shows potent antifungal activity against phytopathogenic fungi.^22–24 However, most studies have only evaluated the antifungal effect of whole essential oil, and the specific active components and mechanisms of action remain unknown. Essential oils containing highly volatile compounds such as monoterpenes, sesquiterpenes and phenyl propionoids are unstable.¹⁴ Moreover, there are more than 133 species of turmeric worldwide,²⁵ and multiple areas of production in China. This instability and wide species and origin variation could affect the quality and antifungal activity of the turmeric essential oil. Therefore, the active components in essential oil will be more practical than the whole essential oil fraction. Chemometrics has become an increasingly useful strategy to uncover bioactive components in herbal medicine. To combine the bioactivity and chemical profiles of herbal extracts with chemometric models such as principal component regression, partial least squares (PLS), multiple linear regression and artificial neural networks, have been utilized to effectively predict these bioactive compounds. PLS analysis is a linear model commonly used to develop relationships between two data sets with few computations and low risk, and it has been applied to predict the bioactive components in natural products.²⁶ A PLS model was used to identify a series of components in pegaga (Centella asiatica) extracts that were responsible for its antioxidant effects.²⁷ The antioxidant property of green tea was predicted from chromatographic fingerprints using PLS and a robust PLS regression.^28,29 The antioxidant property of Radix Puerariae lobatae was also predicted by a PLS model.³⁰ Thus, PLS regression could be an intelligent tool to predict the bioactive compounds from chromatographic profiles.

When treating stored products, fumigation results in fewer residues than direct co-incubation with contaminated commodities, which is important in the elimination of insect pests and the inhibition of fungal growth.^31,32 Herein, we evaluate the fumigant antifungal activities of volatiles from turmeric essential oil against A. flavus growth and aflatoxins production for the first time. Twenty-four samples of turmeric sourced from different areas were evaluated for their antifungal activity with 100 μL essential oils per plate in fumigation. The objective of this study was to develop a PLS model for predicting the fumigant antifungal activity of turmeric essential oil. The antifungal contribution of different components in the turmeric oil was evaluated by the most discriminatory projection in the multivariate space spanned by the PLS model. The results provide a method for screening active compounds from essential oils for use as antifungal agents by fumigation.

2 Materials and methods

2.1 Fungal material

An aflatoxigenic strain of A. flavus was chosen as the test fungus in the present study. A. flavus lyophilized powder (CGMCC 3.4410) was supplied by the China General Microbiological Culture Collection Centre (Beijing, China) and was dissolved in 0.5 mL sterile water (121 °C, 20 min) for culture on Salt Czapek Dox Agar medium (Qingdao Hope Bio-Technology Co., Ltd., Qingdao, China) at constant temperature and relative humidity (28 °C, 90%) for one week. A. flavus conidia suspensions were prepared according to the method described by Christensen et al.³³ Briefly, inoculums suspensions were prepared by scraping off conidia with an inoculation loop, suspending the conidia in approximately 10 mL of sterile distilled water and Tween-20 (1%, v/v), and removing mycelium by gravity filtration through at least three layers of autoclaved cheese cloth. The conidia concentration was determined with a haemocytometer and adjusted to 10⁶ conidia mL⁻¹.

2.2 Plant material and extraction of the essential oil

Twenty-four turmeric samples were purchased from Bozhou herbal market (Anhui, China). These samples were sourced from the Guangxi, Guangdong, Sichuan and Yunnan provinces in China. The herb sample was smashed and filtered through a sieve of 24-mesh. Approximately 50 g of power was weighed and subjected to hydrodistillation for 5 h using a Clevenger type apparatus. The essential oil was collected in a sterilized glass vial and dried over anhydrous sodium sulfate. The anhydrous essential oil was stored at 4 °C in the dark until analysis.

2.3 Antifungal efficacy of turmeric essential oil following fumigation

The relationship between the antifungal efficacy and concentration of the essential oil as a fumigant was evaluated using turmeric from the Sichuan province (SC2), which gave the highest yield of essential oil among the turmeric samples. The volume fraction of turmeric essential oil in each Petri dish for fumigation was measured using the method of Helal et al.³⁴ A. flavus was inoculated into the centre of a Petri plate containing Czapek Dox Agar medium. The desired quantities of essential oil were absorbed onto pieces of sterile filter paper (diameter: 9 cm) under aseptic conditions. The filter papers were adhered to the underside of each Petri plate lid. A series of volumes of essential oil ranging from 50–500 μL per Petri plate were dropped onto filter papers evenly. Parafilm was placed around each Petri plate to seal the joint between the lid and the base. The plates were incubated at 28 °C for 10 days. The antifungal efficacy was evaluated using A. flavus growth, conidia production, and aflatoxin production as detailed in the following sections. The lowest concentration of turmeric essential oil that inhibited 100% of fungal growth was considered the Minimum Inhibitory Concentration (MIC).

The antifungal efficacies of turmeric essential oils from 24 batches with different herbal origins were evaluated at the fumigation concentration of 100 μL essential oil per plate using the above method. Knowledge of the antifungal behaviour could be used to establish chemical component–antifungal effect relationships in subsequent research.

2.4 Inhibition of A. flavus growth

Mycelium growth of the A. flavus strain was evaluated on a Czapek Dox Agar medium. For the fumigation test, sterile filter paper soaked with turmeric essential oil was adhered to the underside of the Petri plate lid. Parafilm was placed around each Petri plate to seal the joint where the lid connected with the base. The plates were incubated at 28 °C for 10 days. Similarly, the Petri plate which was set with a single colony of A. flavus without essential oil treatment was set as the control. All tests were performed in triplicate. At day 10, the radial growth of A. flavus with the essential oil was compared to that in the control and used to calculate the percentage inhibition as follows:³⁵

Percentage mycelia inhibition = [(d_c − d_t)/d_c] × 100%

where d_c is the mean colony diameter for the control and d_t is the mean colony diameter for the treatment.

2.5 Inhibition of conidia production

Fungal conidia production was evaluated using modification of an established method.³⁶ Colonies of A. flavus that had been exposed to turmeric essential oil were incubated for 10 days to allow for conidia formation. Conidia was harvested by adding 5 mL of sterile water containing 1.0% (v/v) Tween-20 to each Petri plate and gently scraping the mycelium surface three times with an inoculation loop to free the conidia. The suspension of conidia was decanted from the Petri plate and centrifuged at room temperature at 5000 rpm for 5 min. The supernatant was discarded and centrifugation was repeated until only 1 mL of a highly concentrated conidia solution remained. The conidia concentration was estimated using a haemocytometer slide (depth 0.1 mm, 1/400 mm²) under a light microscope (IX51, Olympus, Tokyo, Japan).

2.6 Inhibition of aflatoxin production

Aflatoxins in the Petri plates were extracted by ultrasonication for 5 min with 50 mL of methanol/water (80/20, v/v) in an Erlenmeyer flask. The extracts were filtered through filter paper. 10 mL of the filtrate was transferred to another flask and 40 mL 2% Tween-20/PBS was added. After vigorous mixing, the mixed solution was filtered through glass microfiber filter paper and 25 mL of filtrate was passed through an AflaTest-PTM IAC at approximately 1 drop per s. The IAC was then washed twice with 10 mL water at 1 drop per s until some air passed through the column. Aflatoxins were then eluted with 2.0 mL methanol, and evaporated to 0.2 mL under a nitrogen stream at 45 °C. Then the extract was reconstituted with methanol/water (50/50, v/v) to 1 mL and filtered through a 0.22 μm filter. Chromatographic analysis was performed on a Waters Acquity UPLC H-Class system (Waters, MA, USA) equipped with quaternary solvent delivery pump, an auto sampler and fluorescence detector, connected to Waters Empower data software. The chromatographic separation was performed on a Waters Acquity UPLC BEH C18 column (100 mm × 2.1 mm, 1.7 μm) and the column temperature was kept at 30 °C. The mobile phase consisted of methanol/water (50/50, v/v) at a flow rate of 0.2 mL min⁻¹, freshly prepared every day. The injection volume was 1 μL. Aflatoxins were detected at λ_ex 360 nm and λ_em 440 nm. Under optimum conditions, the analysis took around 6.0 min. Aflatoxin inhibition was calculated as follows:

Inhibition (%) = (1 − X/Y) × 100%

where X is the mean concentration of aflatoxin in the treatment and Y is the mean concentration of aflatoxin in the control.

2.7 Static headspace gas chromatography-mass spectrometry (SH-GC-MS) analysis of the turmeric essential oil

In the present study, gas chromatography-mass spectrometry (GC-MS) with headspace sampling³⁷ was applied to detect volatile compounds in turmeric essential oil, to determine exactly what the fumigant components are. Briefly, 1 mL of each turmeric essential oil was placed into 22 mL glass vials sealed with silicone septa and aluminium foil. To mimic the fumigation in fungal culture, vials were then thermostated at 25 ± 0.1 °C. After equilibrium for 24 h, 1 mL of vapour from above the solution was drawn out from the vial using a gas-tight syringe and injected directly into the chromatographic column via a transfer line.

Samples were then analyzed by the SHIMADZU GCMS-QP2010 system (SHIMADZU CO., LTD., Japan). Separation of the components was performed on an InertCap 5MS/NP column (30 m × 0.25 mm i.d., film thickness 0.25 μm). For GC-MS detection, an electron ionization system with ionization energy of 70 eV was used. Helium gas was used as the carrier gas at a constant flow rate of 1 mL min⁻¹. Injector and MS transfer line temperatures were set at 250 °C and 280 °C, respectively. The ion source temperature was 220 °C. The initial column oven temperature was kept at 50 °C for 3 min, and then was gradually increased to 160 °C at a rate of 5 °C min⁻¹ and was held for 5 min at 160 °C, then to 165 °C at a rate of 0.5 °C min⁻¹, then to 280 °C at a rate of 20 °C min⁻¹ and was held for 5 min at 280 °C. The samples were injected manually to the GC-MS system in the split mode with a split ratio of 1 : 10. Components were identified by matching their recorded mass spectra with those of standards from the National Institute of Standards and Technology libraries, and comparison of their retention indices with literature data. The relative percentage content of each component was also calculated by the peak areas as the percentages of total essential oil components. The results were all expressed as a mean ± SD.

2.8 Development of a PLS model

Partial least squares model, a linear regression model, was used to establish the relationship between the bioactive compounds in essential oil and its antifungal activity.²⁷ The chemical compounds in the SH-GC-MS spectrum were used as the X (input) variables and their antifungal activities, including the inhibition of A. flavus growth, conidia growth and aflatoxin production, were used as the Y (output) variables. The validity of the model was assessed using internal cross validation by means of R² and Q² cumulative and permutation tests. In the internal cross validation, R² shows the goodness of fit and Q² shows the goodness of prediction of the model with respect to the Y variables and the significance of the PLS components. For a good model, the value for Q² must be higher than 0.5, and R² must always be larger than Q². The permutation test was used to determine the validity and degree of over-fit for the PLS model. The permutation plot shows the correlation coefficient between the original Y-variable and the permuted Y-variable on the X-axis versus the cumulative R² and Q² on the Y-axis. The results were fitted using a regression line and their intercepts indicated the degree of over-lap. Variable importance plot (VIP) values in the projection were used to analyze the relationship between the X- and Y-variables in the PLS model. VIP summarizes the contribution of each X-variable on the model, and only gives a positive value. According to the principles of variable importance, VIP values larger than 1.0 are the most meaningful.

2.9 Antifungal activity verification by prediction results

To demonstrate whether the compounds predicted above possess antifungal effects, those with the highest contribution to antifungal profiles in the PLS regression model were employed as fumigants to inhibit A. flavus. Using a similar protocol as that described for whole essential oil in Section 2.3, the inhibition on A. flavus growth, conidia production and AFB₁ content acted as the antifungal indexes of screened compounds.

2.10 Statistical analysis

All data are reported as the mean ± standard deviation. All analyses were performed in triplicate. The PLS model was established using SIMCA-P software (v. 11.5, Umetrics, Umea, Sweden).

3 Results and discussion

3.1 Antifungal activities of different volume fractions of turmeric essential oil used in fumigation

Firstly, we established an HPLC method for the simultaneous analysis of four kinds of aflatoxin, i.e., AFG₁, AFG₂, AFB₁, AFB₁ (Fig. 1A). As shown in Fig. 1B, only AFB₁ could be detected in the A. flavus incubation system with high specificity, which is the most frequent occurring mycotoxin with high toxicity. Meanwhile, a reduced AFB₁ concentration in the A. flavus colony treated with 100 μL of turmeric essential oil by fumigation was observed, according to the decreased chromatographic peak area in Fig. 1C. Compared with the control group without any treatment, varying degrees of antifungal activity against A. flavus colonies, conidia production, and AFB₁ production were observed in culture plates containing 50–500 μL of turmeric essential oil after 10 days of incubation. The antifungal efficacy of essential oil fumigants are presented in a dosage-dependant manner (Fig. 2). Although only 10.6% of A. flavus growth was inhibited by 50 μL of essential oil, A. flavus growth, conidia growth, and AFB₁ production were almost stopped after 400 μL of essential oil fumigation. This value was considered as the minimum inhibitory concentration of turmeric essential oil for fumigation of A. flavus. Fig. 3 visually showed the inhibition effects on the growth of A. flavus (CGMCC 3.4410) in the case of 100 μL and 400 μL essential oil fumigation, compared to the untreated control group. No doubt, antifungal effects against multiple types of fungi, including A. flavus, by turmeric essential oil have been reported.²² However, turmeric essential oil was directly applied for incubating contaminated crops in that study, which would be difficult to execute in practice. Actually, fumigation is the most convenient and popular approach for chemical antifungal treatment. This study, for the first time, evaluates the antifungal effects of turmeric essential oil by fumigation. Moreover, we also demonstrated its antifungal activity as a fumigant against both A. flavus (CGMCC 3.4408) and A. flavus (CGMCC 3.4409). As shown in Fig. 4, the fumigation with 400 μL essential oil added for 10 days could drastically eliminate the fungal colony.

Fig. 1 Typical UPLC-FLD chromatograms of (a) mixed standard substance of aflatoxins; (b) aflatoxins produced in the control group without any treatment; (c) aflatoxins produced in the A. flavus colony by fumigation treatment with 100 μL of turmeric essential oil (SC2).

Fig. 2 Antifungal activity of A. flavus (CGMCC 3.4410) growth, conidia production and AFB₁ content at different fumigation concentrations of turmeric essential oil.

Fig. 3 Representative images of inhibition effects against A. flavus (CGMCC 3.4410) growth after fumigant incubation for 10 days. (a) Fungal colony without any treatment. (b) Fungal colony with 100 μL of turmeric essential oil by fumigation. (c) Fungal colony with 400 μL of turmeric essential oil by fumigation.

Fig. 4 Representative fungal colony images of A. flavus (CGMCC 3.4408) (upper) and A. flavus (CGMCC 3.4409) (lower), respectively. Labels (A–C) represent the untreated control group, and those fumigated by 100 μL turmeric essential oil and 400 μL turmeric essential oil for 10 days, respectively.

3.2 Antifungal activity of 24 batches of turmeric essential oil

The antifungal activity of 24 batches of turmeric essential oil samples were compared by adding 100 μL of essential oil. The inhibition effects on A. flavus growth, conidia production, and AFB₁ concentration for the 24 batches of essential oil with different origins are shown in Table 1. Samples possessed varying degrees of inhibition efficacy against A. flavus. For example, for the inhibition of A. flavus growth, the inhibition rate ranged from 7.89% to 47.4%. Even samples with the same origin showed different inhibition behaviours. This variability could be caused by differences in the herb quality between samples. There are a number of factors that can affect the quality of essential oils, including plant species, geographical region of cultivation, harvesting procedure of plants, essential oil extraction methods and storage terms.^38–40 Variation in essential oil quality is exhibited as changes in the chemical components and bioactivity. Thus, although the antifungal activity of turmeric essential oil has been confirmed, this variability can lead to unsteady antifungal activity for whole oil. This is the reason why we here investigate the bioactive compounds with the aim to replace the essential oil in applications.

Table 1 Antifungal activities of 24 batches of turmeric essential oil from different origins (n = 3)

Origin (province)		Inhibition (%)
		A. flavus growth	Conidia production	AFB1 concentration
Sichuan	SC1	28.9 ± 3.58	91.4 ± 1.52	80.7 ± 2.13
	SC2	33.3 ± 3.96	90.3 ± 1.36	82.6 ± 2.18
	SC3	31.1 ± 4.15	59.1 ± 2.69	34.2 ± 3.53
	SC4	26.7 ± 2.79	89.3 ± 2.05	51.9 ± 2.97
	SC5	26.7 ± 3.28	90.1 ± 0.98	66.7 ± 4.08
	SC6	24.4 ± 2.90	65.0 ± 3.69	73.4 ± 3.59
	SC7	26.7 ± 3.85	89.6 ± 1.72	52.3 ± 2.96
	SC8	21.1 ± 4.71	52.2 ± 3.04	81.5 ± 3.06
	SC9	13.2 ± 3.16	54.7 ± 3.84	39.9 ± 4.11
	SC10	34.2 ± 3.88	92.1 ± 1.16	55.1 ± 3.52
Guangxi	GX1	15.6 ± 4.95	27.3 ± 3.84	34.4 ± 3.55
	GX2	26.7 ± 3.68	91.6 ± 0.95	77.5 ± 4.13
	GX3	21.1 ± 3.59	99.5 ± 0.32	95.8 ± 1.89
	GX4	7.89 ± 4.95	81.0 ± 1.89	99.6 ± 0.35
	GX5	34.2 ± 4.10	79.4 ± 2.60	96.6 ± 2.15
Yunnan	YN1	24.4 ± 3.92	30.9 ± 3.37	87.4 ± 2.96
	YN2	10.5 ± 3.57	82.7 ± 3.81	95.0 ± 3.81
	YN3	21.1 ± 3.78	80.4 ± 2.53	97.8 ± 1.08
	YN4	34.2 ± 4.26	84.5 ± 2.96	92.2 ± 2.87
	YN5	21.1 ± 3.17	83.2 ± 2.54	88.7 ± 2.50
Guangdong	GD1	47.4 ± 3.61	98.6 ± 1.06	97.9 ± 1/95
	GD2	34.2 ± 4.85	65.9 ± 2.95	91.2 ± 2.89
	GD3	14.5 ± 4.05	21.9 ± 3.51	82.7 ± 3.75
	GD4	39.5 ± 3.60	77.2 ± 3.05	94.8 ± 2.90

---

3.3 Volatile composition in turmeric essential oil by GC-MS determination

The clear yellow essential oils were extracted from the 24 turmeric samples by hydrodistillation method with the yields ranging from 1.60 to 5.60%. Among the turmeric samples, the herbs from Sichuan gave the highest essential oil yields. The chemical composition of gas from turmeric oil was analyzed using SH-GC-MS, and a representative chromatogram is shown in Fig. 5. Furthermore, we compared the relative ion intensity of samples and three standards (shown in ESI Fig. S1–S3†). From Fig. S1–S3,† three main compounds in essential oil samples, i.e. eucalyptol, betapinene, and camphor, were identified with the similar relative ion intensity as the three chemical standards. In total, 18 peaks were identified in all 24 samples, and these peaks accounted for 99.9% of the volatile constituents. The identified chemical compounds, and their retention times, and their relative percentages are given in Table 2. The majority of the identified compounds were monoterpenes (81.3%) and sesquiterpenes (18.7%). The most abundant compounds were eucalyptol (30.00%), camphor (15.9%) and beta-pinene (17.2%). Compared to our previous work on origin discrimination of turmeric essential oil,⁴¹ these compounds (eucalyptol, camphor, and beta-pinene) represent <1.00% of the whole essential oil. The high volatility of these compounds could be responsible for this result.

Fig. 5 Representative SH-GC-MS chromatograms of volatile gas from turmeric essential oil.

Table 2 Chemical profiles of volatile compounds in turmeric essential oil by SH-GC-MS analysis

Peak no.	Retention time	Name	Relative percentage content (%)
1	7.710	alpha-Pinene	5.97 ± 1.42
2	8.148	Camphene	6.35 ± 1.51
3	9.030	beta-Pinene	17.2 ± 1.35
4	9.868	alpha-Phellandrene	2.55 ± 1.85
5	10.049	3-Carene	0.600 ± 0.550
6	10.500	beta-Cymene	2.78 ± 1.85
7	10.716	Eucalyptol	30.0 ± 0.750
8	14.178	Camphor	15.9 ± 0.940
9	21.127	beta-Elemene	1.38 ± 1.29
10	21.828	alpha-Santalene	1.14 ± 1.04
11	21.847	Caryophyllene	0.760 ± 0.240
12	22.704	alpha-Farnesene	0.210 ± 0.120
13	22.841	alpha-Caryophyllene	0.550 ± 0.520
14	23.386	alpha-Curcumene	6.32 ± 0.630
15	23.698	(−)-Zingiberene	2.57 ± 1.30
16	24.024	alpha-Bisabolol	1.79 ± 1.27
17	24.386	beta-Cedrene	1.52 ± 1.44
18	24.392	beta-Sesquiphellandrene	2.46 ± 1.72

---

3.4 PLS model validation

An internal cross validation was used to estimate the robustness and the predictive ability of the PLS model. The values of R²Y and Q² were as follows: for inhibition of colony growth, 0.822 and 0.518, respectively; for inhibition of conidia growth, 0.936 and 0.773, respectively; and for inhibition of AFB₁ formation, 0.914 and 0.783, respectively. Because the values of Q² were all greater than 0.5, the model is stable and has satisfactory fitness for prediction. A response permutation test for the three first components was used to further validate the PLS model. The results show that the Y-axis intercepts of R² and Q² after 20 permutations were as follows: for inhibition of colony growth, 0.297 and −0.221, respectively; for inhibition of conidia growth, 0.194 and −0.275, respectively; and for inhibition of AFB₁ formation, 0.276 and −0.193, respectively. These results also suggest that the PLS model is valid and does not show over-fit according to the following criteria: Y-axis intercept of R² < 0.3, Y-axis intercept of Q² < 0.05, and R² line not horizontal.²⁷

3.5 Chemical spectrum–antifungal activity relationship based on PLS model

A PLS model was used to evaluate the chemical component–antifungal activity relationship of the essential oil. The results were used to determine the effective compositions of antifungal treatments for A. flavus. The VIP values in the PLS model of the 18 volatile compounds corresponding to each evaluation index are listed in Table 3. These values represent the contribution of each compound to the antifungal activity. Notably, the VIP values of eucalyptol, beta-pinene and camphor were much larger than those of the other compounds, which indicates that these compounds are important contributors to the antifungal activity (inhibition of A. flavus growth, conidia production, and aflatoxins production). As mentioned in Table 2, eucalyptol, beta-pinene, and camphor were the most abundant compounds in the chemical analysis of turmeric essential oil volatiles. The abundance of these compounds and their potent fungal inhibition suggests that they are suitable for the development of novel antifungal agents.

Table 3 VIP values of bioactive compounds based on PLS models

Peak no.	Name	VIP value^a
		A. flavus growth	Conidia amount	AFB1 concentration
a VIP value: variable importance plot (VIP) value is a variable selection method based on the PLS regression obtained by the SIMCA-P software.
7	Eucalyptol	2.71	2.85	2.86
3	beta-Pinene	1.97	1.83	1.94
8	Camphor	1.63	1.55	1.66
14	alpha-Curcumene	1.27	1.25	0.971
2	Camphene	0.783	0.744	0.751
18	beta-Sesquiphellandrene	0.775	0.886	0.697
1	alpha-Pinene	0.693	0.666	0.668
15	(−)-Zingiberene	0.566	0.625	0.687
16	alpha-Bisabolol	0.415	0.380	0.363
10	alpha-Santalene	0.289	0.260	0.270
6	beta-Cymene	0.255	0.192	0.205
9	Beta-Elemene	0.227	0.192	0.227
4	alpha-Phellandrene	0.195	0.174	0.190
5	3-Carene	0.154	0.139	0.143
17	beta-Cedrene	0.147	0.120	0.183
11	Caryophyllene	0.133	0.140	0.0897
12	alpha-Farnesene	0.0937	0.0812	0.0596
13	alpha-Caryophyllene	0.0782	0.108	0.125

---

Based on the PLS model, a loadings plot was constructed to demonstrate the contribution of each chemical variable towards the antifungal activity. As shown in Fig. 6, the 18 identified peaks (compounds) were scattered in the loading plot according to their contribution. According to the principles of the loading plot interpretation, it could be concluded that the outliers of peak 7 (eucalyptol), peak 3 (beta-pinene) and peak 8 (camphor) are the most important contributors to the antifungal activity. Although the antifungal efficiency of these three compounds has been not investigated by ourselves, abundant previous studies have confirmed their desirable antifungal activities.^42–44 Specifically, eucalyptol has shown toxic effects on in vitro mycelium growth of ten different species of mycotoxigenic fungi involved in several plant diseases, but with low phytotoxic effect on plant seed germination.⁴⁵ Pinene, widely distributed in essential oil, showed fungicidal activity to destroy cellular integrity.⁴⁶ The present study showed that PLS analysis would be useful to predict and screen out the bioactive compounds in complex constituents towards antifungal activity. According to the results, eucalyptol, beta-pinene, and camphor will be the most effective compounds from turmeric essential oil for fumigation against A. flavus.

Fig. 6 PLS score scatter plot (w*c[1]/w*c[2]) to represent the contribution of volatile compounds on (a) growth inhibition; (b) conidia inhibition; (c) aflatoxins inhibition. The points of “var_x” represent the 18 identified compounds in volatile gas from turmeric essential oil. The distance between “var_x” point and the axes cross point represents the contribution of chemical compounds.

3.6 Verification of the antifungal activity of the predicted compounds

Three compounds in essential oil, i.e., eucalyptol, beta-pinene and camphor, were respectively utilized as fumigants. Briefly, various concentrations of compounds, specifically 10, 50, or 100 μL of beta-pinene and eucalyptol per Petri plates, and 10, 50, or 100 mg of camphor dissolved in 500 μL ethanol, were dropped onto filter paper adhered to a Petri plate. After 10 days of incubation, the inhibition efficacy on fungal growth, conidia production and AFB₁ production is shown in Fig. 7. As evidenced, all of these three compounds exhibited dose-dependent antifungal activities. Particularly, beta-pinene and eucalyptol presented considerably potent efficiency on the concentration of 100 μL dosage per plate, in which over 50% inhibition efficacy was achieved. As expected, three main compounds provided by PLS model prediction possess fantastic antifungal effects, which demonstrated the feasibility to predict bioactive components by chemometrics.

Fig. 7 Antifungal activities on mycelium growth (a), conidia production (b), AFB₁ content (c) by the fumigation of beta-pinene, eucalyptol, and camphor after 10 days of incubation.

4 Conclusions

This work demonstrates a promising computer-assisted approach, i.e., the PLS model, to screen the potential bioactive compounds as fumigants in turmeric essential oil to effectively inhibit the growth of A. flavus and aflatoxin production. Based on the volatile chemical profile determination of turmeric essential oil and antifungal activities by fumigation, a PLS model was established to predict three of the compounds, i.e., eucalyptol, beta-pinene and camphor, as the most potential antifungal compounds. Both the previous reports and antifungal effect verification results confirmed the applicability of these three compounds as antifungal agents. Thus, this approach would benefit work aiming to figure out promising natural antifungal compounds in plant essential oils more easily.

Acknowledgements

The authors gratefully acknowledge financial supports from the Special Fund for TCM Scientific Research (no. 201407003), the National Major Scientific and Technological Special Project for “Significant New Drugs Development” during the Twelfth Five-year Plan Period (no. 2014ZX09304307-002), the National Natural Science Foundation of China (no. 81274072) and the Xihe New Star Project.

References

1. A. Bouakline, C. Lacroix, N. Roux, J. P. Gangneux and F. Derouin, J. Clin. Microbiol., 2000, 38, 4272–4273.
2. P. J. Cotty and R. Jaime-Garcia, Int. J. Food Microbiol., 2007, 119, 109–115.
3. H. Hitokoto, S. Morozumi, T. Wauke, S. Sakai and H. Kurata, Appl. Environ. Microbiol., 1978, 36, 252–256.
4. A. Bugno, A. A. B. Almodovar, T. C. Pereira, T. D. J. A. Pinto and M. Sabino, Braz. J. Microbiol., 2006, 37, 47–51.
5. M. A. Klich, Mol. Plant Pathol., 2007, 8, 713–722.
6. S. Rawal, J. E. Kim and R. Coulombe Jr, Res. Vet. Sci., 2010, 89, 325–331.
7. J. Nguefack, J. B. Dongmo, C. D. Dakole, V. Leth, H. F. Vismer, J. Torp, E. F. Guemdjom, M. Mbeffo, O. Tamgue, D. Fotio, P. H. Zollo and A. E. Nkengfack, Int. J. Food Microbiol., 2009, 131, 151–156.
8. C. F. Jelinek, A. Pohland and G. Wood, J.- Assoc. Off. Anal. Chem., 1989, 72, 223–230.
9. F. S. Yeh, C. Y. Mimi, C. C. Mo, S. Luo, M. J. Tong and B. E. Henderson, Cancer Res., 1989, 49, 2506–2509.
10. P. A. Murphy, S. Hendrich, C. Landgren and C. M. Bryant, J. Food Sci., 2006, 71, R51–R65.
11. Y. N. Yin, L. Y. Yan, J. H. Jiang and Z. H. Ma, J. Zhejiang Univ., Sci., B, 2008, 9, 787–792.
12. J. Varga, S. Kocsube, Z. Peteri, C. Vagvolgyi and B. Toth, Toxins, 2010, 2, 1718–1750.
13. M. Scordino, L. Sabatino, P. Traulo, G. Gagliano, M. Gargano, V. Pantò and G. L. Gambino, Eur. Food Res. Technol., 2008, 227, 1339–1347.
14. S. Burt, Int. J. Food Microbiol., 2004, 94, 223–253.
15. L. Jing, Z. Lei, L. Li, R. Xie, W. Xi, Y. Guan, L. Sumner and Z. Zhou, J. Agric. Food Chem., 2014, 62, 3011–3033.
16. N. C. Cardenas-Ortega, M. A. Zavala-Sanchez, J. R. Aguirre-Rivera, C. Perez-Gonzalez and S. Perez-Gutierrez, J. Agric. Food Chem., 2005, 53, 4347–4349.
17. A. Kedia, B. Prakash, P. K. Mishra and N. K. Dubey, Int. J. Food Microbiol., 2014, 168–169, 1–7.
18. D. Kitic, D. Pavlovic and S. Brankovic, Curr. Top. Med. Chem., 2013, 13, 2767–2790.
19. F. Bakkali, S. Averbeck, D. Averbeck and M. Idaomar, Food Chem. Toxicol., 2008, 46, 446–475.
20. M. M. G. Yamamoto-Ribeiro, R. Grespan, C. Y. Kohiyama, F. D. Ferreira, S. A. G. Mossini, E. L. Silva, B. A. D. Abreu Filho, J. M. G. Mikcha and J. M. Machinski, Food chem., 2013, 141, 3147–3152.
21. K. Bairwa, J. Grover, M. Kaniaa and S. M. Jachak, RSC Adv., 2014, 4, 13946–13978.
22. O. D. Dhingra, G. N. Jham, R. C. Barcelos, F. A. Mendonça and I. Ghiviriga, J. Essent. Oil Res., 2007, 19, 387–391.
23. A. Apisariyakul, N. Vanittanakom and D. Buddhasukh, J. Ethnopharmacol., 1995, 49, 163–169.
24. S. Sindhu, B. Chempakam, N. Leela and R. Suseela Bhai, Food Chem. Toxicol., 2011, 49, 1188–1192.
25. S. Prasad and B. B. Aggarwal, Herbal Medicine: Biomolecular and Clinical Aspects, 2011, ch. 13.
26. Y. Yan, Q. Wang, W. Li, Z. Zhao, X. Yuan, Y. Huang and Y. Duan, RSC Adv., 2014, 4, 25430–25439.
27. Maulidiani, F. Abas, A. Khatib, M. Shitan, K. Shaari and N. H. Lajis, Food Res. Int., 2013, 54, 852–860.
28. A. M. van Nederkassel, M. Daszykowski, D. L. Massart and Y. Vander Heyden, J. Chromatogr. A, 2005, 1096, 177–186.
29. M. Daszykowski, Y. Vander Heyden and B. Walczak, J. Chromatogr. A, 2007, 1176, 12–18.
30. F. T. Chau, H. Y. Chan, C. Y. Cheung, C. J. Xu, Y. Liang and O. M. Kvalheim, Anal. Chem., 2009, 81, 7217–7225.
31. S. Rajendran and V. Sriranjini, J. Stored Prod. Res., 2008, 44, 126–135.
32. N. Paster, M. Menasherov, U. Ravid and B. Juven, J. Food Prot., 1995, 58, 81–85.
33. S. Christensen, E. Borrego, W. B. Shim, T. Isakeit and M. Kolomiets, J. Visualized Exp., 2012, 62, e3727.
34. G. A. Helal, M. M. Sarhan, A. N. Abu Shahla and E. K. Abou El-Khair, J. Basic Microbiol., 2007, 47, 5–15.
35. R. Shukla, A. Kumar, P. Singh and N. K. Dubey, Int. J. Food Microbiol., 2009, 135, 165–170.
36. N. G. Tzortzakis and C. D. Economakis, Innovative Food Sci. Emerging Technol., 2007, 8, 253–258.
37. A. Ciobanu, I. Mallard, D. Landy, G. Brabie, D. Nistor and S. Fourmentin, Food Chem., 2013, 138, 291–297.
38. A. Rezaei Nejad and A. Ismaili, J. Sci. Food Agric., 2014, 94, 905–910.
39. S. Garg, R. Bansal, M. Gupta and S. Kumar, Flavour Fragrance J., 1999, 14, 315–318.
40. N. Sangwan, A. Farooqi, F. Shabih and R. Sangwan, Plant Growth Regul., 2001, 34, 3–21.
41. Y. Hu, W. Kong, X. Yang, L. Xie, J. Wen and M. Yang, J. Sep. Sci., 2014, 37, 404–411.
42. E. Horvathova, J. Navarova, E. Galova, A. Sevcovicova, L. Chodakova, Z. Snahnicanova, M. Melusova, K. Kozics and D. Slamenova, J. Agric. Food Chem., 2014, 62, 6632–6639.
43. S. M. Seo, J. Kim, J. Kang, S. H. Koh, Y. J. Ahn, K. S. Kang and I. K. Park, Pestic. Biochem. Physiol., 2014, 113, 55–61.
44. A. Frankova, M. Marounek, V. Mozrova, J. Weber, P. Kloucek and D. Lukesova, Foodborne Pathog. Dis., 2014, 11, 795–797.
45. C. Morcia, M. Malnati and V. Terzi, Food Addit. Contam. Part A-Chem., 2012, 29, 415–422.
46. A. C. da Silva, P. M. Lopes, M. M. de Azevedo, D. C. Costa and C. S. Alviano, Molecules, 2012, 17, 6305–6316.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra01725d

‡ Yichen Hu and Jiaoyang Luo contributed equally to this work.

---