Analysis of the Effect of 60Co-γ Irradiation Sterilization Technology on the Chemical Composition of Saffron Using UPLC and UPLC/Q-TOF-MS

To evaluate the effect of 60Co-γ irradiation sterilization technology on the chemical composition of saffron, we collected 10 batches of saffron samples and treated them with different irradiation doses. UPLC characteristic chromatogram showed that there was no significant effect of irradiation on 13 common peak areas. The results of cluster analysis and principal component analysis showed that there were no differences in the chemical composition in nonirradiated and irradiated samples. UPLC/Q-TOF-MS identified 40 characteristic components of saffron, and the results showed that all of these were detected in the saffron samples both with and without irradiation. Irradiation doses at or below 10 kGy had no significant effect on the chemical components of saffron. This provides a sound basis for the use of 60Co-γ ray irradiation sterilization technology during the preparation of original powder saffron as a medicinal herb, for the effective destruction of mycotoxin contamination.


Introduction
Mold is a major cause of deterioration during the harvest and storage of food and medicinal plants. It is relatively easy to avoid this type of contamination because molds can be easily seen on such products. However, contamination with mycotoxins, the toxic metabolite products of molds, is hidden because of the microscopic size of the particles concerned. When food is contaminated with mycotoxins, the sensory characteristics of the food generally do not change signi cantly, so contamination can easily pass unnoticed. Furthermore, mycotoxins are heat resistant; therefore, cooking is not an e ective method for their removal [1]. According to the published literature, mycotoxins have been shown to be carcinogenic [2], mutagenic, teratogenic, and toxic to cells [3]. E ect of irradiation on chemical composition and some other aspects of some medicinal herbs has been studied [4][5][6][7][8][9]. However, the speci c application of irradiation sterilization technology on medicinal herbs needs to be considered based on the characteristics of each species.
Cobalt 60 ( 60 Co-c) irradiation technology is an e ective method for the decontamination of food and medicines from both insect infestation and mycotoxin contamination. It works by destroying the microbes in a sample and by damaging the DNA material in living organisms. Under the correct conditions, gamma irradiation can e ectively destroy mycotoxin, so it is an important technology for plants decontamination [10]. To date, the Codex Alimentarius Commission (CAC) has permitted the use of irradiation treatment for the decontamination of plant-derived food materials in more than 55 countries including the United States, the European Union, South Korea, and China. An irradiation dose below 10 kGy is permitted in this context [11][12][13].
Sa ron (Iridaceae) reportedly has certain positive health e ects, such as activating blood ow and removing blood stasis, cooling the blood detoxi cation, removing depression, and anchoring the mind [14]. It is widely used to promote good blood circulation, to relieve blood stasis [15] and to treat cardiovascular disease [16]. In Chinese medicinal preparations, sa ron is often used in powder form (such as in Ershiwuwei Coral pills, Ershiwuwei Songshi pills, and Dingkun pills). Irradiation is widely used in the manufacture of traditional Chinese medicines for the control of mycotoxin contamination, including in sa ron preparations. However, its e ect on the quality and chemical composition of sa ron is not well understood and demands research. Besides, chemometric methods were often employed to identify di erences between batches [17][18][19]. In this study, we collected 10 batches of sa ron samples and treated them with di erent irradiation doses to evaluate the e ect on chemical composition of the samples. We used UPLC coupled with cluster analysis and principal component analysis (PCA) to evaluate the results qualitatively and semiquantitatively. UPLC/Q-TOF-MS was used for the identi cation of the chemical characteristic components of the sa ron samples with and without irradiation.

Sample Preparation.
For each sample to be analyzed, 200 mg of the powder of sa ron was accurately weighed and mixed with 50 ml of 60% methanol solution. After weighing the resulting solution, it was subjected to ultrasonic treatment for 30 min (at a power of 250 W, frequency of 40 kHz, and temperature of 25°C), cooled, and then reweighed. Make up the less weight with methanol. After ltration, the primary ltrate was discarded and the secondary ltrate obtained from ltration through a 0.22 μm microporous membrane was retained for chromatography analysis. 10.21 mg Crocin I was accurately weighed and placed in a 100 ml volumetric ask, dissolved in methanol and diluted to volume. From this, 2 ml was transferred into a 10 ml volumetric ask and diluted with methanol to volume. e ask was shaken, and the solution was then ltered through a 0.22 μm microporous membrane. e resulting reference standard solution was 20.42 μg/ml in concentration. 5.63 mg Crocin II was accurately weighed and placed in a 100 ml volumetric ask, dissolved in methanol and diluted to volume. From this, 3 ml was transferred into a 10 ml volumetric ask and diluted with methanol to volume. e resulting reference standard solution was 16.89 μg/ml in concentration.

Irradiation.
e sa ron samples were packaged in plastic bottles. Subsamples of all 10 sa ron samples were analyzed chromatographically without irradiation (control group), followed by analysis of 10 further subsamples after being exposed to 60 Co-c irradiation doses of 10 kGy for 5 h or 10 kGy for 10 h in Xi'an Beilin Pharmaceutical Co., Ltd. (China). e absorbed dose was determined with a silver dichromate dosimeter.

Data Analysis
Paired tests were performed to compare the chromatography peak areas (of the two common peaks) with and without irradiation of the samples. e identity and composition of the individual chemical components in the sa ron samples were determined from the chromatography results, with and without irradiation.
Using the characteristic peak area data produced by the UPLC analyses, PCA was conducted to distinguish between the sa ron samples according to their treatment (irradiated versus nonirradiated). Unsupervised PCA was conducted with the peak area of each identi ed chemical component as the X variable and with batch number as the Y variable. e di erence between the chemical composition of the nonirradiated and irradiated samples was further identi ed and analyzed by HCA using SPSS version 20.0 software (SPSS, Inc., Chicago, IL). e 13 characteristic common peak areas and the sample size of di erent groups (the nonirradiated, irradiated, and di erent irradiation doses of sa ron) were used as parameters. e clustering relationship between each sample was calculated using a 33-row and 15-column matrix that was built by SPSS 18.0 statistical software.

Chemical Di erences between Nonirradiated and Irradiated Sa ron Samples.
e qualitative and semiquantitative comparison of the chemical composition of nonirradiated and irradiated sa ron was made through UPLC and Q-TOF-MS analysis. A total of 13 characteristic chromatographic peaks were detected from 10 batches of sa ron (Figure 1,  Supplementary Figure 1). Peak number 8 was Crocin I. e relative retention time of the characteristic chromatograms produced by the sa ron samples was calculated with peak number 8 as the standard peak (S peak) (Supplementary Table 1). e average di erence in the common peak area produced by each sa ron sample with and without irradiation was calculated ( Figure 2). is showed that peaks 6, 7, 8, 9, and 10 all produced a certain degree of reduction after irradiation, with peak number 7 producing more marked decreases than the other peaks. e relative deviations were less than 10% before and after irradiation. e data were tested by paired t test, and bilateral sig value of t test was greater than 0.01. is result suggests that c-irradiation did not have a signi cant e ect on the chemical component composition of the sa ron samples.

HCA and PCA Analysis.
HCA and PCA were used to quantitatively evaluate the diversity of the chemical component di erences between the nonirradiated and irradiated plant samples. In HCA, we used the monitoring method to reveal and compare the connection distance between "different" and "similar" samples, after data normalization processing. In Figure 3, the data relating to the di erent irradiation dose levels (control "nonirradiated"/10 kGy for 5 h/10 kGy for 10 h) from the same batch of sa ron were  Journal of Analytical Methods in Chemistry gathered together to form one group. Under this analysis, nonirradiated, irradiated, and even di erent irradiation doses of the same batch of sample were found to be very close, indicating that the di erence produced by the di erent irradiation doses was small. ere was no signi cant e ect of the irradiation dose on the chemical composition of the sa ron samples. PCA, as an unsupervised pattern recognition method, distinguished the degree of di erence between the irradiated and nonirradiated samples. e data are processed by dimension reduction to carry on PCA, and display of the score and the load for the samples is obtained, with the peak area of each component as the X variable and the sa ron batch number as the Y variable. e 30 sets of chromatography data from the sa ron samples that had been subjected to the 3 di erent irradiation doses (10 samples per dose) were analyzed by PCA. e degree to which the rst two variables explained the total variance among the samples was 93.4%, indicating that the group of sa ron samples was suitable for analysis by PCA. From the loading diagrams for each sample, it was observed that the "distance" between sa ron samples from the same batch that had received di erent irradiation treatments was closer than that between sa ron samples from di erent batches. e di erent treatment groups could not be easily distinguished. is is consistent with the results of the hierarchical cluster analysis, which  Journal of Analytical Methods in Chemistry indicated that there was no signi cant e ect of irradiation treatment on the composition of chemical components in the sa ron samples (Figure 4).

Q-TOF-MS Chemical Composition Identi cation.
In order to further determine the e ects of irradiation on the chemical composition of sa ron, UPLC-ESI-Q-TOF-MS was used to analyze the di erences of Crocin I, II, III, and other characteristic chemical composition between with and without irradiation. e results showed that certain characteristic components including Crocin I, II, III, and other components were all detected in both irradiated and nonirradiated samples. e results can be seen in Table 1 (Supplementary Table 2 and Figure 2). rough the analysis of the mass spectromety fragments, coupled with determination of their accurate molecular weight, as well as their comparison with data from the relevant reference standards, the characteristic components of the sa ron samples were identi ed. A total of 40 components were

Conclusion
is study evaluated the e ect of 60 Co-c irradiation on the chemical composition of sa ron. Two doses of irradiation were tested, up to a maximum of 10 kGy (for 10 h). A variety of statistical methods, including paired t-tests to compare chromatographic peak areas and PCA and HCA, were used to determine di erences between the irradiated and nonirradiated samples. e results indicated that irradiation had no signi cant e ect on the chemical components of sa ron. However, with the increase of irradiation dose and time, there is a trend that the peak area is decreased, so the irradiation dose and time should be strictly controlled. ese ndings provide a sound technical basis for the irradiation of sa ron medicinal herb using original powder using 60 Co-c for the e ective destruction of mycotoxin contamination and indicate that irradiation technology has good application prospects in the sterilization of sa ron and other medicinal plants.

Conflicts of Interest
e authors declare that they have no con icts of interest.