Untargeted metabolomics revealing changes in aroma substances in ﬂ ue - cured tobacco

: The composition and content of aroma sub - stances in ﬂ ue - cured tobacco ( Nicotiana tabacum L. ) will a ﬀ ect the quality of tobacco. To investigate the cor - relation between various aroma substances of K326 before and after ﬂ ue - curing, and their impact on tobacco quality and diversity, this study employed the middle leaves of K326 and KRK26 as test materials. Samples were collected both before and after ﬂ ue - curing for untargeted metabo - lomics analysis. The results of K326 showed that 584 meta - bolites were signi ﬁ cantly di ﬀ erent and there were 44 aroma - related metabolites, including alcohols, aldehydes, phenols, organic acids, etc. The analyzed aroma com - pounds consist of 37 known tobacco aroma substances, while 7 metabolites, previously not associated with tobacco aroma, have been identi ﬁ ed as aroma substances in other food products. These ﬁ ndings suggest that these seven metabolites might may be potential tobacco aroma com - pounds. Further analysis showed that the content of phe - nols, alcohols, and aldehydes increased signi ﬁ cantly after ﬂ ue - curing, but the content of organic acids decreased. Furthermore, the analysis of KRK26 revealed a correlation between the quantity of aroma substances and the type of tobacco. These ﬁ ndings serve as a reference for enhancing the ﬂ ue - curing process of K326 and optimizing the indus - trial production of cigarettes that use cured K326 tobacco leaves.


Introduction
Tobacco (Nicotiana tabacum L.), belonging to the family Solanaceae, is one of the most important industrial crops worldwide and a model plant organism for studying fundamental biological processes [1].In industry, flue-cured tobacco is the main raw material for cigarette production.The composition and content of flue-cured tobacco aroma substances directly determine the quality of cigarette [2], so the study of aroma substances of flue-cured tobacco has important guiding significance for production.The aroma substances of flue-cured tobacco are mainly divided into organic acids, polyphenols, alcohols, aldehydes, ketones, alkaloids, and other categories according to their aroma functional groups [3][4][5], and the contents of them are mainly affected by flue-curing.During the flue-curing process, carbohydrates, proteins, and other organic substances in tobacco are constantly decomposed and transformed, small molecule metabolites continue to generate, and substances related to tobacco aroma are also constantly changing [6].The quantity of aroma substances present in flue-cured tobacco leaves differs significantly from that in fresh tobacco leaves.This variation can impact the accessibility of raw materials and hold considerable significance for production guidance [7,8].The extraction and detection methods of aroma substances in tobacco leaves include solid phase extraction, liquid phase microextraction, and headspace phase extraction [5,9], and liquid chromatography-tandem mass spectrometry (LC-MS/MS) and gas chromatography-mass spectrometry (GC-MS) are commonly used as analytical tools for the absolute quantitative detection of a certain type of chemical composition or several important substances [10,11], such as polyphenols and alkaloids, and lack a comprehensive and systematic description of the complex aroma substances in tobacco leaves [12].Untargeted metabolomics technology can compensate for the limitations of traditional research methods by analyzing all metabolites present in tobacco leaves and characterizing tobacco leaf metabolomics from a global perspective.This approach offers several advantages, including impartiality, comprehensiveness, high efficiency, and convenience [13].Most studies utilizing metabolomics to unveil changes in tobacco metabolites have been restricted to a single detection method (GC-MS or LC-MS).These methods reveal only a limited number and quantity of metabolites, leading to incomplete analysis results.
Metabolomics technology has gained popularity in tobacco research.For example, Liu et al. [14] utilized metabolomics to examine the differences in tobacco aroma precursors during various stages of growth and development, elucidating the alterations in aroma substances, such as polyphenols, organic acids, and fatty acids during the growth process.Li et al. [15] demonstrated the impact of metabolites on enhancing the quality of tobacco leaves and reducing the content of harmful substances during the microbial fermentation process of tobacco.Similarly, using metabolomics, Fu et al. [16] identified the effects of environmental factors on chlorogenic acid, quercetin, malic acid, and other polyphenols and organic acids in tobacco leaves.These studies investigated alterations in tobacco leaf quality, ranging from the growth stage to the fermentation process, using metabolomics.They revealed that it is possible to detect changes in tobacco leaf quality in different modulation processes using metabolomics.However, few reports have explored the quality changes of flue-cured tobacco varieties K326 and KRK26 before and after flue-curing using untargeted metabolomics.
K326 is a high-quality flue-cured tobacco variety widely planted in China, and KRK26 is another variety selected from K326.K326 and KRK26 are popular fluecured tobacco varieties, often used in fen-flavored cigarettes.These varieties contribute to the distinct regional characteristics of these cigarettes.To explore the changes in aroma substances and contents in K326 and KRK26 under the traditional three-stage flue-curing process and compare the differences between the two varieties, this study took K326 and KRK26 as the research objects and carried out untargeted metabolomics analysis of tobacco leaves before and after flue-curing based on LC-MS/MS and GC-MS methods, which will effectively expanded the spectrum of tobacco metabolism, providing more valuable metabolic information.The results of this study will provide a reference for the improvement and optimization of the subsequent tobacco flue-curing process.It will also add some information on metabolites that may affect the flavor of tobacco and provide a reference for the theoretical and applied research on improving the flavor and quality of tobacco leaves.The comparison of the two varieties will also provide a theoretical reference for the selection of raw materials in industrial production.

Research materials
The tobacco varieties used here were K326 and KRK26, which were planted in the research and experimental base of Yuxi City, Yunnan Province.After the tobacco plant grew to maturity, the middle tobacco leaves were harvested and then placed in the bulk curing-barn for fluecuring.The flue-curing process followed the "three-stage" method introduced by Hu et al. [17], with minor adjustments in curing temperature and humidity (Figure 1).The dry-bulb temperature and wet-bulb temperature were closely monitored across the three stages of the process.Fresh tobacco leaves before flue-curing and dried tobacco leaves after flue-curing were collected respectively, and each sample was set up with three replicates.
An Agilent 7890B Gas Chromatograph (Agilent Technologies, USA) and Pegasus BT Mass Spectrometer (LECO, USA) were used to profile the metabolites.GC-MS was performed on a DB-5MS capillary column (30 m × 250 μm i.d., 0.25 unfilm thickness, Agilent J&W Scientific, USA) to separate the derivatives at a constant flow of 1 mL/min helium.One microliter of sample was injected in split mode with a split ratio of 1:10.The injection temperature was 280°C, the transfer line temperature was 320°C, the ion source temperature was 230°C.The temperature rise of the program was followed by initial temperature of 50°C for 0.5 min, 15°C/min rate up to 320°C and staying at 320°C for 9 min.Mass spectrometry was performed using a full scan method with a scan rate of 10 spec/s, the electron energy was −70 V, and the solvent delay was 3 min.

Data analysis
Analysis of data results using principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), and fold change (FC) analysis, differential metabolites were screened according to P-value and FC value (P < 0.05 and FC > 2 or < 0.5).The images were created using the online platform provided by Panomik Biomedical Technology Co. (http://v2.biodeep.cn/dashboard).

Qualitative and PCA results of metabolites in K326
In this study, a total of 1,212 metabolites were identified (919 by LC-MS/MS and 293 by GC-MS, see Tables S1 and  S2).To analyze the data from the fresh leave and dried leave groups in the positive (ESI+) and negative (ESI−) modes of LC-MS/MS analysis and GC-MS analysis, the score plot of the PCA model was generated for the first two principal component analyses (Figure 2).Across all three figures, it was observed that PC1 accounted for more than 70% of the principal components, while PC2 accounted for only about 10%.There was a significant distance between the two groups while clustered together within a group in the PC1 dimension indicating that the difference between the two groups is significant and there was great repeatability of samples within the group.The PLS-DA permutation test diagram (see Figure S1) indicates that the established PLS-DA model is stable and reliable, there is no over-fitting phenomenon, and the data results are available.

Differential metabolites of K326 before and after flue-curing
Different metabolites in the various tobacco samples were identified using P-value and FC analyses were used to identify different metabolites in different tobacco samples (Figure 3).The compounds with P-values <0.05 and FC values >2 or <0.5 were considered those with significant changes.Statistically, a total of 440 differential metabolites were obtained from 919 metabolites obtained by LC-MS/MS.In comparison to the fresh leaf group, the dry leaf group had an increase in the content of 349 metabolites and a decrease in 91 metabolites (Figure 3a).On the other hand, GC-MS analysis detected 293 metabolites, with 144 differential expression metabolites, of which 67 were up-regulated and 77 were down-regulated (Figure 3b).In general, the metabolite content showed an increasing trend after flue-curing.
Untargeted metabolomics revealing changes in aroma substances  3

Aroma substances in K326
Further analysis of differential metabolites resulted in 44 aroma-related metabolites (27 by LC-MS/MS and 17 by GC-MS), including alcohols, aldehydes, phenols, organic acids, etc.Using the substances that have been reported to contribute to the aroma of tobacco or other foods, as a basis, the aroma substances were divided into conventional and potential tobacco aroma substances.The potential aroma substances refer to those that have not been reported in tobacco leaves but are used as flavor substances in some foods.The heat map in Figure 4 illustrates the dynamic changes of 37 conventional and 7 potential tobacco aroma substances before and after flue-curing.Each column represents a tobacco sample, and each row represents an aroma substance.

Metabolites results of KRK26
The types of metabolites detected by KRK26 and K326 were consistent (Tables S3 and S4), but the differential metabolites before and after flue-curing were different.Figure 3c and d illustrates that a total of 624 differential metabolites were identified in KRK26 samples before and after flue-curing, with 465 detected by LC-MS/MS and 161 by GC-MS.Among these metabolites, 49 were aroma substances, including 4 potential aroma substances (Figure 5).These aroma substances in KRK26 were found to include phenols, alcohols, aldehydes, and organic acids, among others.In general, the content of these substances tended to increase after flue-curing, with only scopoletin, kaempferol, quinic acid, 2-hydroxybutanoic acid, oxalic acid, aminomalonate, alpha-ketoisovaleric acid, malonic acid, 2,3-dihydroxybutanoic acid, benzaldehyde, phenylethyl alcohol, and L-rhamnose showing a decrease in content after flue-curing, while the content of other aroma substances increased.

Comparison of aroma substances between K326 and KRK26
The comparison of aroma substances between K326 and KRK26 showed that 28 aroma substances existed in both varieties, and there were significant differences in the content of 16 aroma substances before and after the fluecuring of K326; similarly, 21 aroma substances in KRK26 (Figure 6).After analyzing the content of the 28 common aroma substances in the two varieties after flue-curing, it was found that there were no significant differences except for quercetin 3-O-beta-D-glucosyl-(1→2)-beta-D-glucoside, 2-Isopropylmalic acid, and 4-methylbenzyl alcohol (Table 1).This suggests that the two varieties have similar content of the 28 aroma substances.

Discussion
Currently, over 700 aroma substances have been discovered in burley tobacco, flue-cured tobacco, and oriental tobacco, which mainly comprise organic acids, alcohols, aldehydes, ketones, phenols, esters, alkaloids, and so on [3].However, the current research on tobacco aroma substances tends to concentrate on specific types, which implies certain limitations [18].The present study utilized untargeted metabolomics technology to investigate the alterations in aroma substances before and after flue-curing, providing a comprehensive and unbiased approach to analyzing the intricate aroma substances in tobacco.This technique is highly efficient, convenient and provides a viable method for examining tobacco aroma substances [19,20].
Consistent with previous findings [21 -24], this study revealed a substantial rise in the levels of chlorogenic acid and total polyphenols after flue-curing (as shown in Figure 4), which contributes to the mild, sweet flavor and scent of flue-cured tobacco products.It has been further confirmed that the increase in polyphenol content is due to the pyrolytic conversion of pyrolysates such as lignin and cellulose in tobacco leaves during the curing process [8].Similarly, the content of aldehydes in K326 was increased after flue-curing in this study (as depicted in Figure 4).These findings are in line with previous study [25].Aldehydes play a crucial role in tobacco aroma, as they are closely related to associated with the carbonyl groups present in the molecular structure of tobacco and can enhance contribute to its sweetness [26,27].In addition, similar to the study by Ren et al. [28], the contents of alcohols were increased in K326 after flue-curing.It has also been observed that alcohols contribute to the development of floral and fruity notes in the aromas of tobacco [29,30].All these studies indicate that the content of polyphenols, aldehydes, and alcohols in tobacco leaves generally is increased after flue-curing, which is helpful to improve the quality of tobacco.
Organic acids, including polybasic acid, saturated fatty acid, and unsaturated fatty acid, do indeed have important impacts on the quality and flavor of flue-cured tobacco [15,24].Unsaturated fatty acid can increase the irritation and reduce the smoothness of smoke, while other types of organic acids would increase the mellowness of smoke [31,32].Therefore, a relatively low content of unsaturated fatty acid and a high content of other organic acids in tobacco imply good sensory performance [33].However, previous studies exhibit different varying patterns in the content of organic acids.Zhao et al. [34] found that both saturated and unsaturated fatty acid contents were decreased after flue-curing, while another study showed an opposite results with the contents of both polybasic acid and unsaturated fatty acid increasing after flue-curing [35].Different varying patterns of organic acids might result from different varieties of tobacco, which have been confirmed by other laboratories [36,37].
Our study also showed a decrease in the content of organic  acids but an increase in the content of several unsaturated fatty acids in K326 (Figure 4), whereas the content of all organic acids in KRK26 was increased (Figure 5).Moreover, the variation differences in the content of organic acids might also be related to the growth environment [38,39] and curing and fermentation aging [6,40,41] for flue-cured tobacco.Therefore, it is necessary to adjust the composition and proportion of aroma substances according to the characteristics of the tobacco variety in cigarette production.KRK26 is a high-quality flue-cured tobacco variety selected of K326.In this study, we also compared the differential aroma substances in these two varieties before and after flue-curing.The results revealed that almost half of the aroma substances were present in both varieties (Figure 6), and no significant difference in content was observed for them (Table 1).Besides, the contents of these aroma substances were increased after flue-curing in both varieties, indicating their close genetic relationship.Similar findings have been reported in other studies [42,43], which suggest a certain genetic similarity in the composition and content of aroma substances among relative varieties.These studies further support the idea that the genetic relationship of flue-cured tobacco variety can affect the composition and content of substances in tobacco.In addition, the unique aroma substances of each variety may be an important reason for the differences in flavor between K326 and KRK26, as indicated in some studies [44,45], but the specific mechanisms remain to be further studied.
Untargeted metabolomics is an important approach for identifying potential aroma substances in a global and unbiased manner.In this study, we identified seven potential aroma substances in K326 and four potential aroma substances in KRK26 (Figures 4 and 5), which are known aroma components in various food products.For example, perillyl alcohol is a commonly used natural flavor additive [46], while methyl cinnamate is a safe flavoring agent widely used in the food industry [47].It has also been found to contribute to the flavor of strawberryflavored soybean beverages and apple pomace [48,49].These results suggest that aroma substances not only be related to the flavor of various food products but also play a role in the fragrance of tobacco.However, further research is needed to confirm whether these potential aroma substances are responsible for the aroma of tobacco.
Untargeted metabolomics is a powerful and comprehensive analytical approach that enables the detection of all metabolites in a sample with high efficiency and without bias.In this study, we utilized both LC-MS/MS and GC-MS techniques to expand the metabolite infor-mation and provide a more comprehensive analysis.However, only the relative content of metabolites could be obtained since no standards were added to the samples before untargeted metabolomic assay.Therefore, further improvements are needed for the precise quantitation of the metabolites.

Conclusion
Our untargeted metabolomics study demonstrated that the contents of polyphenols, alcohols, and aldehydes were increased after flue-curing, consistent with previous research findings.However, the content of organic acids was decreased instead, which might be due to the difference of the flue-cured tobacco variety and the growing environment.The results of this study can provide valuable insights for future research on aroma substances in flue-cured tobacco and provides practical theoretical guidance for tobacco production.PLS-DA permutation in GC-MS detection mode.Table S1: The metabolites of K326 before and after flue-curing detected by LC-MS/MS; Table S2: The metabolites of K326 before and after flue-curing detected by GC-MS; Table S3: The metabolites of KRK26 before and after flue-curing detected by LC-MS/MS; Table S4: The metabolites of KRK26 before and after flue-curing detected by GC-MS.

Figure 2 :
Figure 2: PCA score plots derived from LC-MS/MS and GC-MS metabolite profiles of tobacco.(a) Positive ion mode (ESI+) and (b) negative ion mode (ESI−) in LC-MS.(c) GC-MS.Note: in the figures, "XY" represents fresh tobacco leaf samples of K326 and "KH" represents dried tobacco leaf samples after flue-curing of K326.The same is described below.

Figure 3 :
Figure 3: Volcano map of tobacco metabolite expression.(a) The results of K326 by LC-MS/MS and (b) GC-MS and (c) the results of KRK26 by LC-MS/MS and (d) GC-MS.

Figure 4 :
Figure 4: Heat map of aroma substances content change of K326 before and after flue-curing.

Figure 5 :
Figure 5: Heat map of aroma substances content change of KRK26 before and after flue-curing.

Table 1 :
Comparison of the content of 28 aroma substances in K326 and KRK26 of flue-cured tobacco Note: The data in the table represent the average of three replicates for each treatment group.