Determination of volatile organic compounds by HS‐GC‐IMS to detect different stages of Aspergillus flavus infection in Xiang Ling walnut

Abstract The aim of this study was to evaluate the performance of volatile organic compounds (VOCs) for evolution monitoring and early detection of Aspergillus flavus (A. flavus) contamination in walnuts. We successfully applied headspace–gas chromatography–ion mobility spectrometry (HS‐GC‐IMS) to evaluate walnut VOC changes caused by A. flavus contamination. A total of 48 VOCs were identified in walnuts contaminated with A. flavus. After identification of VOCs, a heat map and principal component analysis (PCA) highlighted ethyl acetate‐D, 3‐methyl‐2‐butanol, and cyclohexanone as potential biomarkers specific to A. flavus contamination in walnuts. These results provided valid targets for the development of sensors to evaluate the early mold contamination in stored walnuts.

. When AFs are present in foods at sufficiently high levels, these fungal metabolites can have toxic effects that range from acute (liver or kidney deterioration) to chronic (e.g., liver cancer) toxicity and can be mutagenic and teratogenic (Mahmoud et al., 2014). Thus, the ingestion of foods contaminated with aflatoxins poses a significant threat to human health due to its hepatotoxicity and immunotoxicity (Yang et al., 2015). In consideration of these detrimental properties, further research on the effect of A. flavus on food is important for developing effective strategies to control food safety in foodstuffs.
Walnuts (Juglans regia L.) are an extremely valuable nut species (antioxidant activity, and phenolic and mineral contents of the walnut kernel (Juglans regia L.) as a function of the pellicle color), and they contain the highest amount of PUFAs of edible nuts (Nakanishi et al., 2016). With the development of society, all people begin to pursue diversity, nutritional content, and safety to meet their dietary needs and food preferences for a positive and healthy life (Udomkun et al., 2018). So, walnuts are commonly found in the human diet because of their rich nutrients (Miao et al., 2020;Sánchez-González et al., 2016). Unfortunately, contamination of walnuts by AFs produced by the fungi A. spp. is a serious problem because of their potential threat to health (Amini & Ghoranneviss, 2016). Some studies have investigated the implications of kernel oxidation and fungal growth (A. spp.) that result in the development of carcinogenic aflatoxins in relation to the commercial storage and transportation of walnuts (Campbell et al., 2003). Molds individuals are tiny and hard to detect in the early stages of growth. When the quality of the walnuts is changed to an abnormal state, the damage is irreparable.
The inhibition of fungi before the toxins are produced is more important and is a better strategy than the removal of toxins once produced (Amini & Ghoranneviss, 2016). Therefore, there is an urgent need for a method that can accurately determine the extent of mold growth on walnuts and control this crisis early on. The occurrence of harmful compounds in foodstuffs can result from their mishandling during food production or can be formed during food production, processing, or storage (Hernández-Mesa et al., 2017). Some harmful compounds are also produced in the process of AF contamination in walnuts. Flavor usually determines the overall unique sensory characteristics of food and is also an important tool for evaluating the nutritional value and freshness of food.
The conventional physical and chemical analysis methods for mold detection cannot achieve the requirements of fast and nondestructive testing because of their complex operation steps, time consumption, and poor sensitivity. With the development of chromatographic and spectral technology, the identification of mold in the food industry has started to turn to the detection of substances produced by the growth and metabolism of molds, such as mycotoxins and volatile organic compound biomarkers. Ion mobility spectrometry (IMS) is an instrumental analytical technique of separating the ions of detected substances based on their ion mobility velocity under atmospheric pressure . Headspace-gas chromatography-ion mobility spectrometry (HS-GC-IMS) is a simple, rapid, and sensitive detection technique (Rodríguez-Maecker et al., 2017). This instrumentation combines the outstanding separation capacity of gas chromatography with the advantages of fast response and high sensitivity of ion mobility spectrometry (Gerhardt et al., 2017). This technique has little requirement for sample pretreatment to identify volatile substances in liquid or solid samples (Cavanna et al., 2018). Over the few decades, this technique has been applied in many different research fields for the detection of chemical warfare agents, for security purposes, and for food quality and safety as well as for medical purposes (Jünger et al., 2010). In particular, it has been used for the detection of food-borne microbial spoilers since spoilage of food is often accompanied by the formation of characteristic volatile compounds (Karpas et al., 2002). As a result, the HS-GC-IMS technique can separate and identify volatile compounds in complex matrices, such as aldehydes, ketones, alcohols, amines, and other volatiles. Considering these factors, HS-GC-IMS technology was used to establish an effective method to identify aroma compounds in walnut samples contaminated with A. flavus at different growth stages. Characteristic fingerprint spectra and heat map were used to characterize the infection process of A. flavus, and PCA with cluster analysis was used to explore the utilization of this method for the rapid assessment of the degree of walnut mildew and the feasibility of early warning of the degree of walnut mildew.

| Materials, fungal strains, and inoculum preparation
Unshelled butterfly walnuts (Juglans regia L., Xiangling Variety) were obtained from store for this study and preserved in high barrier bags at −20℃. Aspergillus flavus were a laboratory standard strain NRRL3357, purchased from China General Microbial Culture Collection Centre. This strain forms high concentrations of aflatoxin after growth on YES agar (20 g/L yeast extract, 150 g/L sucrose, 15 g/L agar) at 30°C for 4 days.

| Pretreatment of walnut samples
First, the randomly selected walnuts were peeled and disinfected with 1% sodium hypochlorite. After washing the samples three times with sterile water, the surface of the samples was dried with sterilized filter paper. Then, walnut pulps of the same size, no pests, no mechanical damage were grouped and weighed about 5.5 g per group.

| Preparation of mildew samples
The samples of the treatment groups were inoculated with a concentration of 10 6 /ml A. flavus spore suspension and placed on the water agar medium. Then, they were dried at room temperature and cultured in a 30°C incubator with constant temperature and humidity, and samples not inoculated with A. flavus spores were used as control. At the same time, each group of samples was set up with three parallel groups, a total of 18 groups of samples. Next, the sample changes were closely observed and sampled in a freezer at irregular intervals.

| Statistical analysis
The instrumental analysis software includes LAV (Laboratory Analytical Viewer) and three plug-ins as well as GC × IMS Library Search, which can be used for sample analysis from different angles. The VOC identification was achieved by the National Institute F I G U R E 1 3d topographic and 2D topographic maps for walnut samples with different stages of mold growth. (a) The walnut samples and the 3D topographic plot of walnuts with different stages of mold growth; (b) the 2D topographic plot of walnuts at different times; and (c) the 2D difference spectrum plot of walnuts at different times. W0: walnut samples contaminated by A. flavus for 0 hr; W1: walnut samples contaminated by A. flavus for 12 hr; W2: walnut samples contaminated by A. flavus for 1 day; W3: walnut samples infected by A. flavus for 2 days; W4: walnut samples infected by A. flavus for 4 days; and W5: walnut samples infected by A. flavus for 6 days of Standards and Technology (NIST) reference library (NIST Mass Spectral Library, version 2.0a, 2001) and the comparison of the retention times and mass spectra of authentic standards (Taylor et al., 2017). The spectra were analyzed using the LAV software, and the different profiles and fingerprints of volatile components were constructed using the Reporter and Gallery plug-ins. The PCA and heat map were used for clustering analysis of walnut samples . The heat map and PCA were generated using the R software packages, pheatmap for heat maps, and factoextra for the PCA plots.

| HS-GC-IMS analysis of walnut mold
The differences in volatile compounds in walnut samples with dif-  (Figure 1c). If the VOCs were consistent, the background after deduction was white, while red indicated that the concentration of the substance was higher than in the reference, and blue indicated that the concentration of the substance was lower than in the reference. Most of the signals in the topographic plot of the walnut samples appeared between the retention times of 100 and 450s, and in the infected walnut samples, there were several different signals.
(The retention times were between 350 s and 450 s.) Moreover, the signal intensity was stronger than that observed in the pileus. This may be because the compounds yielding these signals were considered to be weakly polar, considering that nonpolar compounds have a longer retention time on nonpolar columns than polar compounds (Arroyo-Manzanares et al., 2017). After being contaminated with A. flavus, the signals of some compounds (sensitive to temperature and easy to decompose or degrade) disappeared, or the signal intensity decreased (Figure 1c). In contrast, the enhanced intensity of some signals showed that the concentration of some compounds increased after contamination.

| Volatile compound identification in walnut samples at different moldy growth stages
The compounds were characterized by comparing the IMS drift time and retention index with those of the authentic reference compounds. Due to their different concentrations, it was observed that some single compounds might produce multiple signals or spots (dimers or even trimers). A total of 48 typical compounds from the topographic plots were identified with a GC × IMS Library (Figure 2 and Table 1) and are represented by numbers in Figure 2. Furthermore, 15 typical compounds from the topographic plots were not identified as corresponding by names.

| Changes in volatile compounds in walnut samples contaminated by A. flavus
The notable visual plots were chosen and listed together by gallery plot for intuitive comparison. Accordingly, the differences in volatile compounds in walnut samples with different contamination times were observed, and the characteristic fingerprints  corresponding to each stage were established. As shown in Figure

F I G U R E 3 Fingerprint comparison of VOCs in noninoculated samples and A. flavus inoculated samples determined by HS-GC-IMS. Notes:
The darker the spot, the larger is the quantity of volatile compounds. Each row represents all the signal peaks selected in a sample. Each column represents the signal peak of the same VOCs in different samples mold samples, while that of group c were mainly produced during the late-stage mold samples. As is shown in Figure 4, there were many kinds of volatile compounds in the walnut samples, and the signals of these volatile compounds were higher in the late-stage mold samples than in the samples from other periods. However, a significant amount of A. flavus could be seen in the late-stage mold samples ( Figure 1a). Therefore, in order to achieve early monitoring and warning, we focused on the detection targets of these compounds in groups a and d (Figure 4). The compounds in group a ethyl acetate-D, 3-methyl-2-butanol, cyclohexanone, V3, V4, and V15 were strongest in the premold stages, and it can be seen that the signals weakened as growth time increased.

| D ISCUSS I ON
Today, global walnut production is increasing because of the increasing consumer demand for this food. The global production of walnuts is approximately 1,500,000 metric tons, and China, the United States, and Iran are the major producers of walnuts (Amini & Ghoranneviss, 2016  were strongest in the premold stages, and it can be seen that the signals weakened as growth time increased. These results indicated that it is possible to feasible to establish a suitable gas sensor to monitor early mold formation in stored walnuts.
In this study, a simple, specific, and reliable method was de-