1. Introduction
Ozone (O
3) is a highly unstable triatomic molecule which has been shown to have effective germicidal properties. Thus, it has been used in the agri-food industry as an anti-microbial agent to minimise contaminants in raw commodities as it can be lethal to many bacteria and has been used in food packaging applications [
1,
2]. While natural levels of O
3 are between 0.01–0.15 ppm, exposure to higher levels can produce some harmful effects on health, especially respiratory and cardiovascular diseases [
3,
4]. Thus, there are occupational health and safety recommendations including legislative limits for human exposure to O
3 of 0.1 ppm for 8 h continuously or 0.3 ppm for 15 min [
5]. O
3 is also a very reactive gas and will degrade rubber tubing and other materials readily. Thus, for treatment of raw or processed food matrices the materials used for generating and measuring O
3 must not be easily corroded (e.g., polytetrafluoroethylene (PTFE) tubing, glass or stainless steel cylinders for treatment of commodities) or steel storage bins or silos.
Fungal species may have a variable sensitivity/tolerance to gaseous O
3 exposure depending on time of exposure × concentration used [
6,
7,
8]. Sensitivity/tolerance is also influenced by moisture content (m.c.), substrate, and spore morphology [
6]. Inactivation of spores and inhibition of mycelial growth, when exposed to gaseous O
3, have shown variable results. For example, Vijayanandraj et al. [
9] reported that
Aspergillus niger mycelial growth was reduced but that spore germination was unaffected by O
3 treatment. Interestingly, mycelial growth and sporulation were inhibited by continuous low concentration O
3 exposure of peaches when inoculated with
Monilinia fructicola,
Botrytis cinerea,
Mucor piriformis, or
Penicillium expansum and kept in store [
10].
There has been a debate as to whether gaseous O
3 treatment is fungistatic or fungicidal [
11,
12]. Minas et al. [
12] and Hibben and Stotzky [
6] reported that germination rates for thick-walled multicellular spores of
Alternaria and
Stemphylium spp. were the same when exposed to air or 1 ppm O
3. However, germination of thin-walled spores of
Rhizopus stolonifer,
Trichoderma viride,
A. niger at the same O
3 concentration was reduced. O
3 has been shown to act against unsaturated lipids in the microbial cell membranes causing a leakage of their contents, and subsequent microbial lysis [
13,
14].
These studies did not examine whether repair mechanisms may be operative and whether there could be a recovery of germinative capacity subsequently. Mylona et al. [
15] showed that exposure of
Fusarium verticillioides conidia to 200–300 ppm O
3 exposure for 1 h was initially effective but that over the subsequent 10 day period, spore viability recovered and indeed resulted in mycotoxin (fumonisins) production under different a
w conditions. Although the inhibition of spore viability by gaseous O
3 has been examined, the ability for physiological repair probably needs more attention.
Studies with cereals and other food commodities including nuts have suggested that exposure to O
3 gas can indeed influence contaminant mycotoxigenic mould spores and in some cases reduce mycotoxin contamination, with some claims of effective detoxification [
14,
16]. For example, Sultan [
17] reported that the populations of
Aspergillus flavus in stored peanuts were significantly decreased after exposure for 1 h at up to 300 ppm. El-Desouky et al. [
18] showed that aflatoxin B
1 (AFB
1) in wheat was effectively degraded by approximately 55 to 77% after exposure to 40 ppm O
3 for 5 to 20 min. Sahab et al. [
19] found that 40 ppm O
3 for 10 min led to 94.6 and 99.5% reduction in AFB
1 and aflatoxin B
2 (AFB
2) respectively. Chen et al. [
20] suggested that the optimal conditions for detoxification of aflatoxins (AFs) in stored peanuts was by using 6 mg L
−1 (= 2.8 ppm) for 30 min, with degradation rates of the total AFs and AFB
1 being approximately 66%. However, many of these studies were carried out in vitro or did not examine differentially how water availabily will influence the efficacy of the O
3. Mylona et al. [
15] certainly found that maize inoculated with
F. verticillioides and exposed to up to 300 ppm O
3 for 60 min (4 L min
−1) and then stored for 15 and 30 days did provide some control of fumonisins contamination at different storage a
w levels. However, for some commodities, especially those with a high level of fatty acids/lipids, treatment with high concentrations of O
3 can have a negative effect on quality of the food by causing oxidation of lipids, changing colour, modifying some vitamins and phenolic compounds and sometimes causing off-odours [
14,
21]. There have been practically no studies to examine the potential use of gaseous O
3 for the control of colonisation of coffee by ochratoxigenic species and on potential reduction in ochratoxin A (OTA) contamination during storage of coffee beans under different water availability conditions [
22,
23]. In contrast, there have been significant studies to examine the breakdown of other mycotoxins, especially AFs and trichothecenes by treatment with gaseous O
3 in a range of durable commodities, especially temperate cereals [
14].
Thus, the aim of this study was to examine the efficacy of different gaseous O3 concentrations to reduce or inhibit populations of ochratoxigenic fungi and OTA contamination of stored green coffee by the dominant species found in coffee beans (A. westerdijkiae, A. ochraceus, and A. carbonarius) under different aw levels (0.75, 0.90, and 0.95). Studies were carried out using two systems: (a) irradiated green coffee which was inoculated with spores of each individual ochratoxigenic species and exposed to 400 and 600 ppm O3 and then stored for 12 days at 30 °C, and (b) naturally contaminated coffee beans modified to the same three aw levels or that inoculated with a mixture of conidia of these three ochratoxigenic species and exposed to 600 ppm O3 and then stored for 12 days at 30 °C. The effects of treatments on fungal populations 48 h after treatment and after 12 days were determined. The OTA contamination was determined after the 12 day storage period.
2. Materials and Methods
2.1. Fungal Species and Strains Used in These Studies
Strains of two species from the
Aspergillus section
Circumdati group,
Aspergillus westerdijkiae (CBS 121986) and
A. ochraceus (ITAL 14), and one from the Section
Nigri group,
A. carbonarius (ITAL 204) were used in these studies. These were all isolated from green coffee beans and are known as OTA producers [
24,
25]. The CBS 121986 strain was kindly provided to us by Dr B. Patino (Complutense University, Madrid, Spain) and the others by Dr. M.H. Taniwaki (ITAl, Campinas, Brazil).
2.2. Spore Suspensions of the Ochratoxigenic Species for Inoculation of Coffee Beans
The three species were all sub-cultured on 6% green coffee extract and incubated at 25 °C for 7–10 days. 10 mL of sterile water containing 0.01% tween 80 was decanted onto the surface of the cultures. These were agitated with a surface-sterilised bent glass rod to release the conidia. The suspensions were decanted into a 25 mL sterile Universal glass bottle and shaken well. The concentration was quantified with a haemocytometer and diluted with sterile water as required to obtain an inoculum conidial concentration of 104 spores mL−1. This was used for the inoculation of the naturally contaminated coffee beans and for the experiments with indiviudal species and irradiated coffee beans.
2.3. Development of the Moisture Adsoprtion Curves for Natural and Irradiated Green Coffee Beans
Moisture adsorption curves were constructed to determine the amounts of water necessary for addition to naturally contaminated Arabica green coffee beans, or that which had been gamma irradiated (12–15 kGys; Synergy Healt, Swindon, Berkshire, U.K.) [
25]. Known amounts of water were added to 5 g green coffee bean sub-samples and equilibrated at 4 °C for 24 h, then returned to ambient conditions and the water activity (a
w) of the hydrated green coffee beans was measured. This was done using an Aqualab 4TE (Decagon Devices Inc., Pullman, WA, USA). The coffee bean samples were then dried at 110 °C for 24 h and kept in a desiccator at room temperature for 1 h and weighed to determine the moisture content. The moisture adsorption curves differed between the two types of coffee beans. Thus to obtain the target a
w values for natural and irradiated coffee beans for the experiments the amounts of water necessary for 0.75, 0.90, and 0.95 a
w treatments were 0.25, 0.9, 2.1 mL, and 0.7, 1.5, and 2.5 mL per 5 g
−1, respectively.
2.4. Ozone Generation, Measurement, and Exposure of Coffee Treatments
All O3 treatments were carried out in a fume cupboard with an extractor to avoid user exposure when carrying out these experiments. The system used consisted of a corona discharge ozone generator (C-Lasky series, Model CL010DS, AirTree Ozone Technology Co., Ltd., Sijhih City, Taipei County 22150, Taiwan). The O3 generator was was placed outside the fume cupboard and connected via PTFE tubing to the bottom of a glass column via an inlet valve. Up to 50 g of coffee could be placed in the glass column. The glass column was held uprght in a steel laboratory clamp stand in the fume hood. The top of the glass column had an exit bi-valve which was also connected to an O3 analyser (Model UV-100, Eco Sensors Inc., Santa Fe, NM 87505, USA). This was to ensure that the exposure of the coffee bean samples was kept at the target O3 treatment exposure level during the treatment exposure period (400 or 600 ppm O3, 6 L min−1; 60 min; 25 °C) accurately. The rest of the O3 was vented via the fume hood to the exterior. Only Teflon tubing was used to avoid degradation of the components during experiments.
2.5. Effect of Gaseous O3 Treatment on Fungal Populations and Ochratoxin A Contamination of Stored Coffee Beans Inoculated with Each Individual Ochratoxigenic Species
The irradiated Arabica green coffee beans (3 × 150 g for each fungal species) were weighed and placed into sterile 500 mL Pyrex glass bottles (Thermo Fisher Scientific Ltd., Basingstoke, Hampshire, UK). The coffee beans were modified to the target aw levels of 0.75, 0.90, and 0.95, excluding 1 mL for the inoculation with the test fungal species. After equilibration overnight, 1 mL of the individual conidial suspension (104 condia ml−1) was added separately to each set of treatments. The glass bottles were shaken to mix the conidial inoculum with the coffee beans and left for 1 h. Approximately 45 g was introduced into the glass column in the sterile flow bench and then connected to the O3 treatment apparatus which was already set at the target O3 concentration generation level. A total of three replicates of each fungal species treatment were exposed to 400 and 600 ppm O3 for 60 min at a flow rate of 6 L min−1. The control treatments were inoculated and exposed to air only.
After treatment the coffee beans were placed directly into 50 mL surface-sterilised glass containers (Magenta, Sigma-Aldrich Ltd., St Louis, MI, USA) and covered with microporous lids. These were all placed into larger plastic environmental chambers which also contained 2 × 500 mL of a glycerol/water solution in 750 mL glass beakers to maintain the equilibrium relative humidity (ERH) of the atmosphere in the environmental chamber and stored at 30 °C for 12 days.
After 48 h and 12 days, sub-samples were taken and used for quantifying the populations of the inoculated species in relation to the O3 treatments. The fungal populations were quantified by placing 10 g in 90 mL of sterile water and shaking well. After serial dilutions, 200 μL of each dilution were spread plated onto three replicate malt extract agar (Thermo Fisher Scientific Oxoid Ltd., Basingstoke, Hampshire, U.K.) plates at different dilutions and incubated at 30 °C for 4–5 days before enumeration. After 12 days, the remaining treatment and replicate samples were all dried and stored at −20 °C until OTA analyses.
2.6. Effect of O3 on the Total Fungal Populations in Naturally Contaminated Coffee Beans and That Inoculated with a Mixture of the Three Ochratoxigenic Species
2.6.1. Effects of O3 on Fungal Populations in Naturally Contaminated Green Coffee Beans
Naturally contaminated coffee beans were rewetted according to the water absorption curve as described previously to obtain the target aw levels of 0.75, 0.90, and 0.95. Three replicates of each treatment (45 g) were exposed for 60 min to 600 ppm O3 in the glass column system at a flow rate of 6 L min−1. Directly after the treatment with O3, approximately 3 g sub-samples were collected from the top, middle, and bottom of the glass treatment chamber as detailed previously, combined, and used for the determination of the total fungal populations.
The rest of the coffee beans were stored for 12 days in the environmental chambers in separate glass containers with microporus lids and containing glycerol/water solutions (500 mls × 2 in glass beakers) to maintain the ERH of the atmosphere in the closed containers. Again, after 12 days storage the fungal populations were determined using the serial dilution method as described previously.
The rest of the sample was dried and stored at −20 °C for later OTA quantification.
2.6.2. Effect of O3 on Naturally Contaminated Coffee Beans with Additional Mixed Inoculum of the Three Ochratoxigenic Species
The same aw treatments were prepared (0.75, 0.90, 0.95 aw) taking into account the addition of the fungal inoculum. A 5 mL spore suspensions of 104 conidia ml−1 was prepared as described previously of each species (A. westerdijkiae, A. ochraceus, and A. carbonarius). They were mixed together to obtain a total of 15 mL. The three replicates of each rewetted naturally contaminated coffee bean treatments were inoculated with 1.5 mL of the mixed spore inoculum and mixed well for 2 min. The replicate coffee bean treatments (45 g) were then placed in the glass column system and exposed to 600 ppm O3 at a flow rate of 6 L min−1 for 60 min.
The fungal populations were determined after treatment and after storage for 12 days as described previously using serial dilution. The rest of the samples were dried and stored at −20 °C for later OTA quantification.
2.7. Ochratoxin A Extraction and Quantification
The dried coffee samples were milled using a Waring Laboratory homogeniser (model 7009G; (Waring Laboratory Science, Torrington, CT, USA)) for 5 min at maximum speed and 10 g of the milled dried coffee beans extracted with a 50 mL methanol:water (70:30) solution in 1% sodium bicarbonate. The extracts were then filtered and 5 mL diluted with 45 mL phosphate buffered saline (PBS/Tween (0.01%
v/
v) and applied to an immunoaffinity column (Neogen Europe Ltd., Auchincruive, Ayr, Scotland, UK). 1.5 mL was dried and 0.5 mL of acetonitrile:water (50:50) added. The final extracts were analysed by HPLC (Agilent, Berkshire, UK). The retention time of OTA under the conditions described was approximately 2.5 min. The mobile phases used were acetonitrile (57%): acetic acid (2%): water (41%) [
15]. A 20 μL aliquot of the extracted toxin from the treatments and replicates were injected into the HPLC system. The conditions for OTA detection and quantification were as follows:
Mobile Phase Acetonitrile (57%): Water (41%): Acetic acid (2%)
Column 120CC-C18 column (Poroshell 120, length 100 mm, diameter 4.6 mm, particle size 2.7 micron; 600 Bar)
Temperature of column: 25 °C
Excitation: 330 nm
Emission: 460 nm
Flow rate: 1 mL min−l
Volume of sample injected: 20 µL
Retention time: Approximately 2.49 min
Run time: 17 min
Limit of detection: 0.01 ng g−1
Limit of Quantification: 0.039 ng g−1
2.8. Statistical Analyses
Normality of the fungal populations and OTA data for the experimental treatments of aw × O3 concentration (400, 600 ppm, or 600 ppm only) × storage time (48 h, 12 days) was checked using the Kolmogorov-Smirnov test. Analysis of data, the factors and response and their interaction were examined by the Kruskal–Wallis (non-parametric) if the data was not normally distributed. For normally distributed data, the datasets were analyzed using a Minitab 16 package (Minitab Inc., 2010. State College, Centre County, PA, USA). All experiments were carried out in triplcate and repeated once. The statistical significant level was set at p ≤ 0.05 for all single and interacting treatment factors.