A Comparison of Volatile Organic Compounds in Puerariae Lobatae Radix and Puerariae Thomsonii Radix Using Gas Chromatography–Ion Migration Spectrometry

: Puerariae Radix is one of the most widely used ancient traditional Chinese medicines and is also consumed as food, which has rich edible and medicinal value. Puerariae Radix can be divided into Puerariae Lobatae Radix (PL) and Puerariae Thomsonii Radix (PT). These two medicinal materials are very similar, and they are often mixed up or misused. In this study, gas chromatography–ion migration spectrometry (GC-IMS) was used to analyze the volatile organic compounds (VOCs) of PL and PT, and the differences in VOCs were analyzed using ﬁngerprint, principal component analysis (PCA), and orthogonal partial least squares discriminant analysis (OPLS-DA). The results showed that a total of 173 VOCs were obtained from PL and PT, and 149 were qualitatively identiﬁed, including 38 aldehydes, 22 alcohols, 22 ketones, 19 esters, 13 esters, 10 acids, 10 pyrazines, 6 terpenes, 4 furans, and 2 pyridines. The characteristic VOCs of PL and PT were clariﬁed by constructing GC-IMS ﬁngerprints. PL and PT can be effectively distinguished, and ﬁve characteristic VOCs were screened using PCA and OPLS-DA analysis methods. This study identiﬁed and evaluated the types and differences in VOCs in PL and PT. The established method is simple, rapid, accurate, and sensitive, and it provides theoretical guidance for the identiﬁcation, tracing, and quality evaluation of PL and PT.


Introduction
Puerariae Lobatae Radix (PL) and Puerariae Thomsonii Radix (PT) are traditional medicinal materials.They are included in the "List of Items Both Food and Drugs" by the Ministry of Health of China, detailing their widespread use in medicine, health products, food, and so on [1,2].In order to improve their taste, efficacy, and convenience, they are usually eaten in powder form.Although China has formulated systems and regulations for the application of powdered traditional Chinese medicine and food, the application of powdered Chinese medicine faces obstacles.Furthermore, powdered Chinese medicine does not display the obvious characteristics of conventional Chinese medicine, so it is relatively difficult to identify.The authenticity of the powder of traditional Chinese medicine directly affects the safety and efficacy of traditional Chinese medicine, so the scientific identification of the powder of Chinese medicine is very important.PL and PT are derived from the dried roots of leguminous plants Pueraria lobata (Willd.)Ohwi and Puerariathomsonii Benth., respectively.They can release muscles and subside fever, encourage the production of

GC-IMS Analysis of VOCs in PT and PL
GC-IMS was used to analyze the VOCs of PL and PT, and a three-dimensional spectrum was obtained, in which the X axis represents the ion drift time, the Y axis represents the retention time of the gas chromatograph, and the Z axis represents the peak intensity used for quantification, as shown in Figure 1.We can observe the difference in VOCs in the PL and PT samples.To facilitate observation, the following two-dimensional top view is used for comparison.As shown in Figure 2, the horizontal coordinate is ion drift time, the red vertical line at 1.0 is the normalized reactive ion peak (RIP peak), and the vertical coordinate is the retention time of gas chromatography.Each point on either side of the RIP represents a volatile organic compound.The color represents the peak strength of the substance.From blue to red, the darker the color, the greater the peak intensity.There are certain differences in VOCs in PT and PL samples, as can be seen in Figure 2.
In order to further visually compare the differences in VOCs, the spectra of PL samples were selected as the reference, and the spectra of the PT samples were deducted from the reference ratio to obtain the difference comparison diagram of different samples, as shown in Figure 3.If the two volatile substances are consistent, the deducted background is white, while red means that the concentration of the substance is higher than the reference, and blue means that the concentration of the substance is lower than the reference.It is easier to distinguish the difference between two samples using contrast atlases.certain differences in VOCs in PT and PL samples, as can be seen in Figure 2.
In order to further visually compare the differences in VOCs, the spectra of PL samples were selected as the reference, and the spectra of the PT samples were deducted from the reference ratio to obtain the difference comparison diagram of different samples, as shown in Figure 3.If the two volatile substances are consistent, the deducted background is white, while red means that the concentration of the substance is higher than the reference, and blue means that the concentration of the substance is lower than the reference.It is easier to distinguish the difference between two samples using contrast atlases.In order to further visually compare the differences in VOCs, the spectra of PL samples were selected as the reference, and the spectra of the PT samples were deducted from the reference ratio to obtain the difference comparison diagram of different samples, as shown in Figure 3.If the two volatile substances are consistent, the deducted background is white, while red means that the concentration of the substance is higher than the reference, and blue means that the concentration of the substance is lower than the reference.It is easier to distinguish the difference between two samples using contrast atlases.

Fingerprints of VOCs in PL and PT
The construction of characteristic flavor fingerprints of PL and PT can provide an effective means for quality evaluation the and identification of the variety.To find the exact difference in VOCs between samples of PL and PT, the GC-IMS results of the two samples were further analyzed using the Gallery Plot plug-in, and the VOCs detected in each sample were selected for a fingerprint comparison (Figure 4).Each row in the diagram represents all of the selected signal peaks in the sample, and each column represents

Fingerprints of VOCs in PL and PT
The construction of characteristic flavor fingerprints of PL and PT can provide an effective means for quality evaluation the and identification of the variety.To find the exact difference in VOCs between samples of PL and PT, the GC-IMS results of the two samples were further analyzed using the Gallery Plot plug-in, and the VOCs detected in each sample were selected for a fingerprint comparison (Figure 4).Each row in the diagram represents all of the selected signal peaks in the sample, and each column represents the signal peaks of the same volatile organic compound in a different sample.Some substances are followed by _M, D, and T, which are monomers, dimers, and trimers of the same substance, respectively.The numbers are unidentified peaks, and the darker the color of each bright spot, the greater the compound content.The complete volatile information for each sample and the differences in volatiles between the samples are outlined in Figure 4.

Fingerprints of VOCs in PL and PT
The construction of characteristic flavor fingerprints of PL and PT can provide an effective means for quality evaluation the and identification of the variety.To find the exact difference in VOCs between samples of PL and PT, the GC-IMS results of the two samples were further analyzed using the Gallery Plot plug-in, and the VOCs detected in each sample were selected for a fingerprint comparison (Figure 4).Each row in the diagram represents all of the selected signal peaks in the sample, and each column represents the signal peaks of the same volatile organic compound in a different sample.Some substances are followed by _M, D, and T, which are monomers, dimers, and trimers of the same substance, respectively.The numbers are unidentified peaks, and the darker the color of each bright spot, the greater the compound content.The complete volatile information for each sample and the differences in volatiles between the samples are outlined in Figure 4.

Identification of VOCs in Different PL and PT
A total of 173 VOCs were detected from PL and PT using GC-IMS analysis, as shown in Table 1.A total of 149 VOCs (monomers, dimers, or trimers) were identified by comparing the NIST2020 vapor phase retention index database built into the practical Vocal software with the IMS migration time database of G.A.S.Among them, there were 38 aldehydes, 22 alcohols, 22 ketones, 19 esters, 13 terpenes, 10 acids, 10 pyrazines, 6 terpenes, 4 furans, and 2 pyridines.In addition, the peak areas of PL and PT show significant differences in the content of VOCs, such as 3-methylbutyraldehyde, 1-octene-3-ol, e-2-hexene-1-ol, isovalerate leaf alcohol ester, butyl acetate, 2,3-dimethyl-5-ethylpyrazine, 2-hexanone, 1,8-cineole, delta-decenolactone, citronellal, Z-6-nonenal, (E, E)-2,4-decenal, camphor, alpha-terpinol, α-pinene, etc. Principal component analysis (PCA) is a multi-variable data analysis tool that converts and reduces the dimensions of the information collected by the sensor to obtain the most important factor with the largest contribution rate, and it reflects the difference in the test samples on the PCA diagram [33].In order to distinguish the difference between PL and PT, PCA was performed on all samples of PL and PT.As shown in Figure 5, there are clear differences between PL and PT.If the distance between the samples is close then the difference is small.If the distance is long then the difference is obvious.As can be seen from Figure 5, the distance between PL and PT is very long, which means that the VOC contents of them are significantly different.

Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA)
PCA focuses on describing the classification trend of samples.Unlike PCA analysis, OPLS-DA is a supervised analysis that can statistically analyze complex data dimensionality reduction, visualize the data, and then build a model to predict the data.In order to further explore and judge the differences and accuracy of VOCs in PL and PT, we further evaluated the feasibility of GC-IMS technology for rapid authenticity identification.The peak volume of 149 VOCs with large differences in selected content was taken as a variable, and the OPLS-DA scores were obtained with partial least squares discriminant analysis.The results are shown in Figure 6, which are consistent with the results of PCA, and different Pueraria samples are clearly distinguished.According to the data processed by SIMCA software, the model can relatively accurately summarize, explain, and predict; the VOC composition of PL and PT is identifiable according to this study; and different varieties can be distinguished to clarify the differences between PL and PT. Figure 6 shows the verification of the OPLS-DA model by using permutation testing.It can be seen from Figure 7 that R 2 intersects the vertical axis (0, 0.842), Q 2 intersects the vertical axis (0, 0.0186), and the slope of the two regression lines is large.It was confirmed that the model could be used to study the classification and discrimination of VOCs in two different varieties of PL and PT via verification.
The variable importance projection (VIP) of the OPLS-DA model with different peak volumes of VOCs is highlighted in Figure 8.The larger the VIP value, the more significant the difference.By observing the VIP value, potential markers can be analyzed.The results showed that there were five VOCs with a VIP value > 1 and p < 0.05, including 2-methyl-3-furanthiol, 1-propanol, ethyl acetate, gamma-butyrolactone-M, and methyl hexanoate-D.The above five VOCs are important indicators for the classification and identification of PL and PT, and they can provide a reference for the rapid authenticity identification of the two pieces.PCA focuses on describing the classification trend of samples.Unlike PCA analysis, OPLS-DA is a supervised analysis that can statistically analyze complex data dimensionality reduction, visualize the data, and then build a model to predict the data.In order to further explore and judge the differences and accuracy of VOCs in PL and PT, we further evaluated the feasibility of GC-IMS technology for rapid authenticity identification.The peak volume of 149 VOCs with large differences in selected content was taken as a variable, and the OPLS-DA scores were obtained with partial least squares discriminant analysis.The results are shown in Figure 6, which are consistent with the results of PCA, and different Pueraria samples are clearly distinguished.According to the data processed by SIMCA software, the model can relatively accurately summarize, explain, and predict; the VOC composition of PL and PT is identifiable according to this study; and different varieties can be distinguished to clarify the differences between PL and PT. Figure 6 shows the verification of the OPLS-DA model by using permutation testing.It can be seen from Figure 7 that R 2 intersects the vertical axis (0, 0.842), Q 2 intersects the vertical axis (0, 0.0186), and the slope of the two regression lines is large.It was confirmed that the model could be used to study the classification and discrimination of VOCs in two different varieties of PL and PT via verification.The variable importance projection (VIP) of the OPLS-DA model with different peak volumes of VOCs is highlighted in Figure 8.The larger the VIP value, the more significant the difference.By observing the VIP value, potential markers can be analyzed.The results showed that there were five VOCs with a VIP value > 1 and p < 0.05, including 2-methyl-3furanthiol, 1-propanol, ethyl acetate, gamma-butyrolactone-M, and methyl hexanoate-D.The above five VOCs are important indicators for the classification and identification of PL and PT, and they can provide a reference for the rapid authenticity identification of the two pieces.

Discussion
In this study, we used GC-IMS for the first time to analyze and identify VOCs in PL and PT.A total of 173 VOCs were detected, and 149 of them were identified, mainly including aldehydes, alcohols, ketones, lipids, and other components, by rapidly comparing the types and contents of VOCs in PL and PT by observing the size and color changes in the sample points representing compound information.By constructing GC-IMS fingerprints, it was shown that the VOCs of PL and PT have extremely high similarity, but the content differences between the groups are obvious.Using principal component analysis and partial least squares discrimination, the distribution of VOCs of PL and PT samples occupies a relatively independent space in the diagram, which can be easily distinguished.Then, the VIP value and p-value were used to identify five different markers of PL and PT, which provided a scientific basis for rapid identification.Compared with traditional analytical methods such as enthrone colorimetry and high-performance liquid chromatography for the identification of PL and PT, GC-IMS technology has great room for the development of identifying the origin of Chinese medicinal materials and counterfeit and shoddy materials.Not only can the composition differences in VOCs of Chinese medicinal materials be analyzed, but samples with similar compositions of VOCs can also be accurately classified according to the content differences in characteristic volatile substances.The experimental results of this study show that GC-IMS can effectively analyze and identify the VOCs in PL and PT, detect the difference between PL and PT, and reach scientific judgments.Moreover, this method requires less sample dosage and is simple in the process of drug pretreatment, which has great application potential, and it provides a scientific basis for the research and development of PT and PL identification in the future.

Conclusions
The rapid identification of traditional Chinese medicines based on "odor" information is an important part of the traditional identification method of traditional Chinese medicines [34].For example, Houttuynia Cordata has a strong fishy smell, and Xiangjiapi has a special fragrance.Experienced pharmacists can directly and quickly identify authenticity and even evaluate quality based on the unique smell and odor thickness of traditional Chinese medicine.With its fast and convenient advantages, up until now, this method has spread as a traditional identification approach.However, for some decoction pieces with insufficient odor information or even weak odor, it may be difficult to quickly realize the identification of traditional Chinese medicine using the traditional "sniffing" method.As a trace detection technology for VOCs, GC-IMS technology cleverly combines the advantages of the rapid identification of traditional traits with the accuracy and quantification of modern instrument analysis.It can be used to quickly and accurately detect information on VOCs in traditional Chinese medicine to allow the inheritance and development of traditional skills.At present, this technology is widely used in food, agriculture, medicine, and other fields.It is mainly used for the rapid detection and characterization of VOCs in samples, as well as the comparative analysis of the differences in VOCs in different samples, and many studies have shown that GC-IMS technology can be used for the identification or classification of two/multiple types of samples [35,36].

Figure 1 .
Figure 1.GC-IMS three-dimensional spectrum of PL and PT.

Figure 2 .
Figure 2. GC-IMS two-dimensional spectrum of PL and PT.

Figure 1 .
Figure 1.GC-IMS three-dimensional spectrum of PL and PT.

Figure 1 .
Figure 1.GC-IMS three-dimensional spectrum of PL and PT.

Figure 2 .
Figure 2. GC-IMS two-dimensional spectrum of PL and PT.

Figure 2 .
Figure 2. GC-IMS two-dimensional spectrum of PL and PT.

Figure 3 .
Figure 3.Comparison of GC-IMS differences between PL and PT.

Figure 3 .
Figure 3.Comparison of GC-IMS differences between PL and PT.

Figure 4 .
Figure 4. Fingerprints of VOCs in PL and PT.

Figure 4 .
Figure 4. Fingerprints of VOCs in PL and PT.

Figure 5 .
Figure 5. PCA analysis of PL and PT.

Figure 6 .
Figure 6.OPLS-DA analysis of VOCs in PT and PL.Figure 6. OPLS-DA analysis of VOCs in PT and PL.

Figure 6 .
Figure 6.OPLS-DA analysis of VOCs in PT and PL.Figure 6. OPLS-DA analysis of VOCs in PT and PL.

Figure 6 .
Figure 6.OPLS-DA analysis of VOCs in PT and PL.

Figure 7 .
Figure 7. Permutation test results of VOCs in PT and PL.

Figure 8 .
Figure 8. VIP value of the characteristic variables.

Figure 7 .
Figure 7. Permutation test results of VOCs in PT and PL.

Figure 6 .
Figure 6.OPLS-DA analysis of VOCs in PT and PL.

Figure 7 .
Figure 7. Permutation test results of VOCs in PT and PL.

Figure 8 .
Figure 8. VIP value of the characteristic variables.Figure 8. VIP value of the characteristic variables.

Figure 8 .
Figure 8. VIP value of the characteristic variables.Figure 8. VIP value of the characteristic variables.

Table 1 .
Results of VOC analysis of PL and PT.