Comparative metabolomic of cigar tobacco leaves: A preliminary investigation of potential discriminatory metabolites between Yunnan and Foreign

Abstract

A systematic understanding of the metabolic components in cigar tobacco leaves (CTLs) from different ecological regions remains limited. This study aimed to clarify the substance differences between CTLs from Yunnan and major foreign production areas and to identify potential discriminant metabolites. A total of 25 CTL samples were collected from four production areas in Yunnan and five different foreign countries. The metabolic profiles were comprehensively characterized using continuous flow analysis, ion chromatography, GC-MS, and UPLC-MS/MS. The results revealed that Yunnan CTLs were characterized by a lower sugar-nicotine ratio and significantly higher contents of starch, pectin, and lignin compared to foreign samples. Phenolic acids, alkaloids, flavonoids, amino acids and derivatives, terpenoids, and lipids were identified as the primary components across all CTLs. A total of 398 differential metabolites (248 volatile and 150 non-volatile) were identified, predominantly from the classes of amino acids and derivatives, alkaloids, terpenoids, and lipids. From these, 12 metabolites were selected as potential candidates for distinguishing CTL origin. KEGG enrichment and MetOrigin analyses indicated that the differential metabolites were mainly involved in the biosynthesis of alkaloids, amino acids, flavonoids, lipids, and terpenes, with these pathways being significantly influenced by microorganism and enzyme-mediated carbon and nitrogen coupling metabolism. This exploratory study provides a metabolic foundation for improving production techniques and for the analysis of style-defining substances in Yunnan CTLs.

Graphical abstract

Highlights

• •A total of 12 potential biomarkers were identified for distinguishing Yunnan and Foreign cigar tobacco leaves.
• •The KEGG enrichment and MetOrigin analysis showed that the differential metabolites were mainly involved in the biosynthesis of alkaloids, amino acids, flavonoids, lipids and terpenes.
• •The carbon and nitrogen coupling metabolism mediated by microorganisms and enzymes influenced the composition and content of cigar tobacco leaves.

1. Introduction

The production of high-quality cigar tobacco leaves (CTLs) is highly dependent on specific ecological conditions, with climate, water, and soil being the pivotal factors shaping their style and characteristics. Consequently, the world's high-quality cigar tobacco production areas are only distributed between the Tropic of cancer and the tropic of capricorn, especially around latitude 23 °N; the main foreign CTLs production areas include Cuba, the Dominican Republic, Indonesia, Nicaragua, Brazil, Mexico, the United States, Honduras, Ecuador, the Philippines, Jamaica, and Cameroon, etc [1,2]. Currently, the main producing areas of CTLs in China are Hainan, Sichuan, Hubei, and Yunnan, while Yunnan is mainly distributed in Pu'er, Lincang, Dehong, and Yuxi. The results from the cigar manufacturing industry and consumer feedback both indicate that the Chinese cigar tobacco leaves have prominent characteristics and broad application prospects. However, compared to foreign cigar leaf, there are still sensory quality defects such as slightly weaker aroma characteristics and taste characteristics, slightly poorer aroma richness, slightly poorer smoke mellowness and sweetness [3].

Since sensory quality is fundamentally determined by chemical composition, understanding these underlying substances is crucial. Recent studies have begun to shed light on the differences in aroma components, key odorants, and sensory characteristics of CTLs from various regions [[3], [4], [5]], laying a foundational data framework for analyzing CTL material composition. Despite this progress, a significant knowledge gap remains, there is still a lack of systematic, comprehensive identification and comparison of the global metabolic profiles of cigar tobacco produced under different ecological conditions. Most previous research has focused on specific compound classes or sensory attributes, leaving the broader metabolome largely unexplored.

Metabolomics has been widely applied to study various small molecules or metabolites in cells, tissues, or organisms, enabling comprehensive characterization of small molecules present in biological samples. Complex and high-throughput analysis platforms such as gas chromatography (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) are commonly used, greatly facilitating our understanding of global metabolomics and metabolic pathway networks [6]. Therefore, the central hypothesis of this study is that CTLs from Yunnan and major foreign production areas possess distinct metabolic fingerprints resulting from their different growing environments, and that these differences can be systematically identified to reveal the chemical basis for their divergent sensory qualities.

The objective of this study was to uncover the global metabolome in CTLs samples using unbiased profiling and metabolic fingerprinting methods to evaluate metabolites of multiple chemical classes associated with different pathways and to identify key differential compounds and potential discriminant metabolites between Yunnan and foreign CTLs. The novelty of this work lies in its integrative and comprehensive nature, combining conventional chemical analysis with both volatile and non-volatile metabolomic profiling, and further tracing the potential microbial origins of characteristic metabolites. We anticipate that this study will provide a crucial chemical baseline and data reference for optimizing cultivation, air-curing, and fermentation processes for Yunnan CTLs, ultimately contributing to the analysis and enhancement of their quality style characteristics.

2. Materials and methods

2.1. CTLs samples

This study employed a total of 25 commercially sourced, fermented cigar tobacco leaf (CTL) samples. These samples were strategically selected to represent four major producing areas within Yunnan Province, China (Pu'er, Lincang, Dehong, and Yuxi), and five prominent foreign production areas (Cuba, Dominican Republic, Indonesia, Brazil, and Mexico). The selection aimed to capture the diversity of CTLs available on the market from these key regions.

To ensure sample comparability, all leaves were selected from a similar stalk position and had undergone a standardized post-harvest process, including air-curing and primary fermentation. The main variation among the samples lies in their commercial types (wrapper, binder, filler) and quality grades, reflecting real-world product diversity. Three independent biological replicates were prepared from different batches of leaves for analysis. Detailed information regarding the specific cultivar and exact harvest season (year) for each sample is comprehensively listed in Supplementary Table S1. It is important to note that variations in these agronomic and post-harvest factors, inherent to their different geographical and production origins, could contribute to the observed metabolic differences.

2.2. Determination of chemical compositions

The content of each index was determined by using a continuous flow analytical system or ion chromatography [7]. The contents of total sugar, reducing sugar, magnesium, polyphenol, total nitrogen, total nicotine, chlorine, potassium, pectin, starch, pH, petroleum ether extract, xylogen were separately determined in accordance with the standards in the tobacco industry YC/T 159–2019、YC/T 161–2002、YC/T 249–2008、YC/T 468–2013、YC/T 217–2007、YC/T 162–2011、YC/T 175–2003、YC/T176-2003、YC/T 346–2010、YC/T 347–2010、YC/T 222–2007.

2.3. Determination of volatile aroma compounds

The indices of 44 volatile aroma compounds were detected by steam distillation extraction and combined gas chromatography/mass spectrometry (GC-MS) [8,9]. Quantitation based on the quantitation ion which is unique in the coeluted components increased the accuracy of peak integration.

2.4. UPLC-ESI-MS/MS analysis

Dissolve 50 mg of lyophilized powder with 1.2 mL 70 % methanol solution, vortex 30 s every 30 min for 6 times in total. Following centrifugation at 12000 rpm for 3 min, the extracts were filtrated before UPLC-MS/MS analysis. The extracts were analyzed using an UPLC–ESI– MS/MS system (UPLC, ExionLC™ AD, https://sciex.com.cn/; MS, Applied Biosystems 6500 Q TRAP, https://sciex.com.cn/).

The analytical conditions were as follows, UPLC: column, Agilent SB-C18. Mobile phase: Phase A is ultrapure water (with 0.1 % formic acid), Phase B is acetonitrile (with 0.1 % formic acid). Gradient program: The proportion of phase B was 5 % at 0 min. Within 9.00 min, the proportion of phase B linearly increased to 95 % and remained at 95 % for 1min. From 10 to 11.1 min, the proportion of phase B decreased to 5 % and balanced at 5 % until 14 min. The ESI source operation parameters were as follows: source temperature 500 °C: ion spray voltage (IS) 5500 V (positive ion mode)/-4500 V (negative ion mode), and oteher operation parameters were performed according to previously reported [1]. A specific set of MRM transitions were monitored for each period according to the metabolites eluted within this period.

2.5. HS-SPME-GC-MS analysis

Chromatography and mass spectrometry were performed with the assistance of Wuhan Metware Biotechnology Co., Ltd. (Wuhan, China). Volatile metabolites in CTLs were analyzed by HS-SPME-GC-MS. A total of 1.5 g CTLs powder was placed in a 10 ml glass vial and extracted by headspace solid phase microextraction (120 μm DVB/CAR/PDMS fiber, Supelco, Bellefonte, USA) at 60 °C for 15 min. After extraction, volatile metabolites were identified according to previously reported [10]. The MS was operated in selected ion monitoring (SIM) mode for the identification and quantification of metabolites.

2.6. Statistical analysis and visualization

The mass spectrometry data were processed using the software Analyst 1.6.3. The chromatographic peaks of the mass spectrometry data were integrated and corrected using the MultiOuant software. Principal component analysis (PCA), hierarchical cluster analysis (HCA), orthogonal partial least square discriminant analysis (OPLS-DA) and KEGG enrichment analysis were used to analyze non-volatile and volatile differential metabolites. The screening criteria for differential metabolites were VIP >1, P value < 0.05, Fold change≥2. The trace analysis of characteristic metabolites was performed by MetOringin [11], Simole MetOrigin analysis (SMOA) model was selected, and common tobacco (Nicotiana tabacum) was selected as host. ROC curve analysis was performed with pROC 1.16 in R software (version 3.5.1) to screen for potential discriminatory metabolites.

Two-way ANOVA was used to analyze the difference of chemical composition indexes between Foreign and Yunnan. Fisher's LSD was used to compare the level of difference in variables between groups. A probability of P < 0.05 indicated that the differences were significant. The SPSS 24.0 (IBM) was used for descriptive statistical analysis and analysis of variance. OriginPro 2024b (10.1.5.132) was used to perform hierarchical cluster analysis and draw polar heat maps.

3. Results

3.1. Conventional chemical composition

In addition to total nitrogen, chlorine and potassium, there were significant differences in total sugar, reducing sugar and nicotine between Yunnan and foreign countries (among groups), while there were only significant differences in pH and polyphenol among 9 different producing unit (within groups) (Table 1). The results indicated that the differences between groups were significantly greater than the differences within groups, and the grouping Settings were reasonable. The results of the two-way ANOVA test showed that the contents of total sugar, reducing sugar, magnesium and polyphenol in Yunnan CTLs were significantly lower than those in Foreign CTLs. The contents of total nicotine, pH, pectin, starch, lignin and petroleum ether extract were significantly higher than those of CTLs from Foreign. There were no significant differences in total nitrogen, potassium and chlorine (Table 1). The nicotine, reducing sugar and pectin were selected as potential discriminatory metabolites of conventional chemical components, which had great influence on quality.

Table 1.

Two-way ANOVA of conventional chemical components of cigar tobacco leaves between Foreign and Yunnan.

Origins	Producing unit	Total sugar	Reducing sugar	Total nitrogen	Total nicotine	Chlorine	Potassium	Magnesium	Pectin	Starch	pH	petroleum ether extract	Xylogen	Polyphenol
Foreign	ID	0.40 ± 0.05	0.33 ± 0.07	3.88 ± 0.34	1.22 ± 0.17	1.14 ± 0.21	4.14 ± 0.03	1.36 ± 0.04	4.37 ± 0.59	0.71 ± 0.29	6.08 ± 0.08	3.49 ± 0.71	1.59 ± 0.08	3.78 ± 0.55
	BR	0.34 ± 0.07	0.31 ± 0.07	3.87 ± 0.28	1.47 ± 0.5	0.74 ± 0.21	4.84 ± 0.62	1.3 ± 0.13	4.93 ± 0.67	0.70 ± 0.06	6.45 ± 0.12	3.24 ± 0.37	2.13 ± 0.25	3.72 ± 0.28
	DO	0.40 ± 0.06	0.35 ± 0.05	4.05 ± 0.32	4.50 ± 0.58	1.58 ± 0.95	3.79 ± 0.89	0.89 ± 0.24	4.69 ± 0.47	1.08 ± 0.24	6.06 ± 0.46	5.89 ± 0.06	1.56 ± 0.3	4.48 ± 1.58
	MEX	0.41	0.31	4.77	3.1	0.27	4.71	1.06	4.2	0.45	6.01	1.95	1.83	3.42
	CU	0.38 ± 0.1	0.22 ± 0.21	3.54 ± 0.11	2.07 ± 0.04	1.04 ± 0.43	4.24 ± 0.18	0.91 ± 0.25	3.85 ± 0.57	0.97 ± 0.51	6.38 ± 0.11	4.24 ± 0.26	1.76 ± 0.34	3.82 ± 0.67
Yunnan	PE	0.30 ± 0.06	0.1 ± 0.04	4.04 ± 0.44	5.12 ± 1.2	0.74 ± 0.05	4.9 ± 0.19	0.61 ± 0.08	6.13 ± 0.89	1.10 ± 0.23	6.83 ± 0.1	5.78 ± 0.14	2.24 ± 0.59	2.99 ± 0.31
	LC	0.37 ± 0.05	0.18 ± 0.01	4.63 ± 0.25	3.07 ± 0.39	0.41 ± 0.18	5.89 ± 0.22	0.82 ± 0.1	6.24 ± 0.51	1.19 ± 0.1	6.10 ± 0.1	3.66 ± 0.69	1.83 ± 0.24	3.88 ± 0.5
	DH	0.34 ± 0.02	0.08 ± 0.01	3.95 ± 0.56	3.53 ± 0.84	2.65 ± 0.53	4.58 ± 0.81	1.24 ± 0.09	7.63 ± 0.9	1.31 ± 0.25	6.61 ± 0.15	4.57 ± 0.2	2.20 ± 0.3	3.23 ± 0.24
	YX	0.27 ± 0.05	0.07 ± 0.03	3.97 ± 0.19	3.23 ± 0.13	1.14 ± 0.23	3.83 ± 0.03	0.88 ± 0.05	4.95 ± 0.49	1.16 ± 0.06	6.76 ± 0.1	6.99 ± 0.5	3.43 ± 0.21	2.54 ± 0.15
Two-way ANOVA P-Values
Origins		0.018	<0.0001	0.169	0.011	0.57	0.090	0.039	<0.0001	0.003	0.007	0.031	0.005	0.015
Producing unit		0.37	0.74	0.85	0.83	0.06	0.07	0.25	0.57	0.84	0.012	0.34	0.90	0.014
Origins × Producing unit		0.43	0.82	0.56	0.80	0.20	0.29	0.60	0.58	0.61	0.018	0.29	0.91	0.049

Note: The unit of each index in the table is %, P ≤ 0.05. PE, LC, DH, YX are respectively represented as Pu'er, Lincang, Dehong, and Yuxi. CU, DO, ID, BR, MEX, are respectively represented as Cuba, Dominican Republic, Indonesia, Brazil, and Mexico.

3.2. Volatile aroma compound

Through quantitative analysis of 44 volatile aroma components detected in foreign and Yunnan CTLs, the results showed that 15 aroma components such as valeraldehyde, pyridine and megastigmatrienone were significantly higher in Yunnan, while 29 aroma components such as 1-Penten-3-one, 2,3-Pentanedione and β-cyclocitral were significantly higher in Foreign. The results of hierarchical cluster analysis showed that 44 volatile aroma components were divided into 2 categories, which could distinguish Yunnan CTLs from Foreign CTLs. Further by OPLS-DA by variable importance projection (VIP) > 1 substancehas 18 kinds, in which pyridine and beta-damascone and megastigmatrienone content is higher in Yunnan CTLs. The contents of 1-penten-3-one; 2,3-pentanedione; 3-methyl-2-butena; n-hexanal; 2-acetyl-1; pyrroline; β-cyclocitral; indole; 6,8-nonadien-2-one,8-methyl-5-(1-methylethyl)-, (6e); cyclohexanone, 2,3-dimethyl-2-(3-oxobutyl); beta-ionone 5,6-epoxide and 2(4h)-benzofuranone,5,6,7,7a-tetrahydro-4,4,7a-trimethyl were relatively high in Foreign CTLs. Secondly, the types of odorant substances with relatively high content in CTLs abroad were significantly more than those in Yunnan (Fig. 1). According to the clustering results, pyridine, megastigmatrienone and β-cyclocitral with VIP value > 1.6 and relative content difference were selected as the characteristic aroma substances to distinguish Foreign and Yunnan CTLs.

Fig. 1.

Polar Heatmap of volatile aroma compounds in cigar tobacco leaves. Chemical names in red font indicate VIP >1.

3.3. Volatile and non-volatile metabolic profiles

3.3.1. Metabolic profiling

A total of 1105 non-volatile metabolites were identified in the 25 CTLs samples, of which 541 were annotated by KEGG. The Pearson correlation coefficient R2>95 %, which met the requirements for subsequent analysis. Ten classes of metabolites were detected in all CTLs, including phenolic acids (16.56 %), alkaloids (14.12 %), flavonoids (14.03 %), amino acids and derivatives (12.58 %), lipids (12.04 %), and organic acids (10.41 %) (Fig. 2A). Accounted for 79.76 % of total tissue metabolites.

Fig. 2.

The composition of non-volatile (A) and volatile (B) metabolites in cigar tobacco leaves.

A total of 615 volatile metabolites were detected based on GC-MS, 223 of which were annotated to the KEGG pathway. The detected metabolites were divided into 15 categories, of which terpenoids (22.44 %), lipids (14.63 %), heterocyclic compounds (14.15 %), ketones (9.92 %) and hydrocarbons (8.94 %) accounted for more than 70 % (Fig. 2B).

The non-volatile metabolites PCA analysis showed that samples in the same group could not only be clustered together, but also separated from samples in other groups, and the trend of metabolite separation among groups was obvious (Fig. 3A). Importantly, the CTLs of YNLC and YNYX were more similar, while the clustering of YNDH and YNPE was similar. For foreign samples, except BR-3, the other origin units are close to each other. Similarly, the HCA analysis showed that samples from foreign also clustered together (Fig. 3B). These results indicate that the non-volatile metabolites of CTLs from foreign countries and Yunnan are significantly different.

Fig. 3.

Multivariate statistical analysis of metabolites in cigar tobacco leaves (PCA, HCA). 1105 non-volatile metabolites were analyzed by PCA (A) and HCA (B). PCA (C) and HCA (D) analysis of 615 volatile metabolites.

The PCA analysis of volatile metabolites in the two groups showed a weak separation trend between the groups (Fig. 3C). The HCA analysis results showed that the CTLs samples were grouped into four categories, among which YNDH, YNPE and YNLC were grouped into one category, Cuba, Mexico, Indonesia and Brazil were grouped into one category, and YNYX and Dominica were grouped separately into one category. One sample from Indonesia is similar to that from Dominica (Fig. 3D). In conclusion, there are some differences in the volatile metabolites of CTLs produced under different ecological conditions, but they have the general characteristics of convergence.

3.3.2. Differential metabolites

The OPLS-DA was used for quantitative and modeling analysis of metabolites, the score map was drawn, and VIP values were obtained. R²Y represents the interpretation rate of the model, Q² is used to evaluate the prediction ability of the model, and R²Y > Q² indicates that the model is well established. According to cross-validation, the interpretation of PC1 and PC2 of non-volatile metabolites to the dataset was 23.2 % and 9.01 %, respectively, indicating large differences between groups and small differences between samples within groups. The prediction model R²Y = 0.989, Q² = 0.953, R²Y > Q² (Fig. 4A and B). The interpretation of PC1 and PC2 of volatile metabolites were 23.9 % and 16.3 %, respectively, and the prediction model R²Y = 0.977, Q² = 0.945, and R²Y > Q² (Fig. 4C and D). The results show that the two metabolomic models are well established and have good predictive ability and interpretation effect for CTLs from Yunnan and Foreign.

Fig. 4.

Analysis of volatile (A, B) and non-volatile (C, D) metabolites OPLS-DA in cigar tobacco leaves.

We further selected VIP >1, Log₂FC > 2 and P < 0.5 as differential metabolites, and Fig. 5 shows the relative content difference and statistical significance of metabolites in Foreign vs Yunnan. A total of 248 volatile differential metabolites were identified, of which 88 were significantly down-regulated, the “N, N′-diferuloylputrescine, retusin and 5-hydroxyl-3′,4′,7,8-tetramethoxyl flavone N, N' -diferuloylutamide” were the top3 differentially varied substances. And 160 were significantly up-regulated, and the “tomatine, N-(3-indolylacetyl)-l-alanine and 4,4,5-trihydroxy-l, l'-di-2-propenylbiphenyl” were the three substances with the largest Log₂FC (Fig. 5A).

Fig. 5.

Volcanic plots of nonvolatile (A) and volatile (B) differential metabolites of Foreign vs Yunnan.

A total of 150 volatile metabolites were identified, of which 107 were significantly down-regulated, “p-menth-8-en-3-ol-acetate, benzoic acid, 1-methylethyl ester and phenol, 3,5-dimethyl” were the Top3, and 43 were significantly up-regulated. “bornyl acetate, (R) −3, 7-dimethyl-6-octenol and styrene were the largest differential multiples (Fig. 5B).

The differential metabolites were classified and counted based on substance classification. The results showed that the primary classification of non-volatile differential metabolites mainly belonged to amino acids and their derivatives (28 %), alkaloids (18.2 %), phenolic acids (11.9 %), lipids (11.9 %) and organic acids (10.2 %) (Fig. 6A). The alkaloids were mainly alkaloids, phenolamines and indole alkaloids (12.6 %), and the lipids were mainly free fatty acids (7.6 %). The main volatile differential metabolites were terpenoids (19.5 %), lipids (14.8 %), heterocyclic compounds (13.40 %) and aromatics (10.1 %) (Fig. 6B). In summary, the main metabolites of CTLs from Foreign vs Yunnan were amino acids and their derivatives, alkaloids, terpenoids and lipids.

Fig. 6.

Classification and pie chart of nonvolatile (A) and volatile (B) differential metabolites of Foreign vs Yunnan.

3.3.3. Metabolic pathways of differential metabolites

KEGG pathway enrichment analysis showed that non-volatile differential metabolites were significantly enriched in 11 metabolic pathways (P < 0.05), including “aminoacyl-tRNA biosynthesis, tropinane, piperidine and pyridine alkaloid biosynthesis, tryptophan metabolism, biosynthesis of amino acid, d-amino acid metabolism, glucosinolide biosynthesis, arginine biosynthesis” (Fig. 7A). The volatile differential metabolites are mainly involved in “biosynthesis of various plant secondary metabolites, monoterpenoid biosynthesis and phenylalanine metabolism” (Fig. 7B).

Fig. 7.

KEGG enrichment diagram of nonvolatile (A) and volatile (B) differential metabolites of Foreign vs Yunnan.

3.4. Characteristic metabolites and potential markers

ROC curve analysis is widely recognized as an objective and effective method to evaluate biomarker performance. To identify the characteristic metabolites of CTLs in Yunnan, strict screening criteria were applied: AUC = 1.0, Log₂FC > 2, VIP >1, P < 0.01. These criteria enabled us to identify the characteristic metabolites more accurately. For non-volatile metabolites, a total of 11 metabolites with distinctive characteristics were identified, mainly steroid alkaloids, amino acids and their derivatives, lignans, phenolic acids, free fatty acids and coumarins (Table 2). Based on ROC curves and AUC values, we suggest that tomatine, n-(3-indolyl) -l-alanine, and 5,6,7-trimethoxycoumarin may be potential non-volatile discriminatory metabolites.

Table 2.

The potential non-volatile discriminatory metabolites of cigar tobacco leaves between Foreign and Yunnan. The red font indicates the differential metabolites with variable importance projection (VIP) > 1.

Metabolites	AUC	VIP	P-value	Log2FC	Formula	Class II
Tomatidine	1	2.06	3.27E-05	17.89	C27H45NO2	Steroid alkaloids
N-(3-Indolylacetyl)-l-alanine	1	2.06	0.000111	17.77	C13H14N2O3	Amino acids and derivatives
4,4,5-Trihydroxy-l,l'-di-2-propenylbiphenyl	1	2.05	0.004481	17.52	C18H18O3	Lignans
l-γ-Glutamyl-l-leucine	1	1.89	0.000522	4.98	C11H20N2O5	Amino acids and derivatives
L-Prolyl-l-Leucine	1	1.12	3.2E-06	3.28	C11H20N2O3	Amino acids and derivatives
Dihydrocaffeic acid	1	1.40	1.4E-08	−2.30	C9H10O4	Phenolic acids
12,13-DHOME; (9Z)-12,13- Dihydroxyoctadec-9-enoic acid	1	1.93	8.82E-06	−2.32	C18H34O4	Free fatty acids
Hexadecanedioic acid	1	1.90	3.05E-05	−2.40	C16H30O4	Free fatty acids
9,16-Dihydroxypalmitic acid	1	1.83	2.6E-05	−2.58	C16H32O4	Free fatty acids
9-Hydroperoxy-9Z,11E-Octadecadienoic Acid	1	2.00	3.24E-07	−2.83	C18H32O4	Free fatty acids
Benzoyltartaric acid	1	1.48	1.86E-08	−2.91	C11H10O7	Phenolic acids
7S,8S-DiHODE; (9Z,12Z)-(7S,8S)- Dihydroxyoctadeca-9,12-dienoic acid	1	2.00	3.1E-07	−2.92	C18H32O4	Free fatty acids
10,16-Dihydroxypalmitic acid	1	1.79	0.000178	−2.95	C16H32O4	Free fatty acids
5,6,7-Trimethoxycoumarin	1	1.78	0.002976	−5.18	C12H12O5	Coumarins

As for volatile metabolites, a total 10 characteristic compounds were identified, belonging to terpenes, aromatics, heterocyclic compounds, hydrocarbons, ethers and lipids (Table 3). The bornyl acetate, 5,6,7, 8-tetrahydroquinoxaline and benzoic acid, 1-methylethyl ester could be defined as potential discriminatory metabolitesfor volatile metabolites.

Table 3.

The potential volatile discriminatory metabolites of cigar tobacco leaves between Foreign and Yunnan.

Metabolites	AUC	VIP	P-value	Log2FC	Formula	Class I
Bornyl acetate	1	1.95	1.62E-06	12.68	C12H20O2	Terpenoids
Octen-1-ol, 3,7-dimethyl-, (R)-	1	1.59	7.41E-06	3.79	C10H20O	Terpenoids
Ethylbenzene	1	1.83	1.15E-07	3.00	C8H10	Aromatics
Ethanone, 1-(3,5-dimethylpyrazinyl)-	1	1.80	1.15E-07	2.92	C8H10N2O	Heterocyclic compound
Benzene, 1,3-dimethyl-	1	1.81	1.15E-07	2.10	C8H10	Aromatics
Trimethyltridecane	1	1.90	1.15E-07	−2.32	C16H34	Hydrocarbons
Benzene, 1-methoxy-4-(1-methylpropyl)-	1	1.46	1.15E-07	−3.29	C11H16O	Ether
5,6,7,8-Tetrahydroquinoxaline	1	1.96	1.15E-07	−3.40	C8H10N2	Heterocyclic compound
Benzoic acid, 1-methylethyl ester	1	1.97	1.15E-07	−3.77	C10H12O2	Ester
p-Menth-8-en-3-ol, acetate	1	1.50	1.15E-07	−5.24	C12H20O2	Ester

3.5. MetOrigin analysis of the differential metabolites

MetOrigin analyses were performed for 11 non-volatile and 10 volatile characteristic metabolites, using sankey networks to integrate statistical and biological associations. The results showed that non-volatile characteristic metabolites were involved in the “degradation of flavonoids”, “linoleic acid metabolism”, and “the biosynthesis of cutin, xylocin, and wax”, “suberine and wax biosynthesis”. The volatile metabolites are enriched into two metabolic pathways “ethylbenzene degradation” and “xylene degradation”.

Further identification of the microbial communities related to non-volatile metabolites showed that the microorganisms closely related to flavonoid degradation were mainly Pseudomonadota and Bacillota. The phloretin hydrolase was involved in degradation of flavonoid. The 3-Hydroxyphloretin is a metabolic substrate, while dihydrocaffeicacid and phloroglucinol are metabolites (Fig. 8A). Associated with linoleic acid metabolism are the microorganisms of Podospora and Pyricularia, and linoleate lipoxygenase and hydroperoxidase isomerase are involved in linoleic acid metabolism (Fig. 8B).

Fig. 8.

BIO-Sankey network diagram of characteristic metabolite traceability analysis.

The microbial communities associated with xylene degradation, a volatile characteristic metabolite, were Novosphingobium, Sphingobium, Pseudoxanthomonas, Rhodococcus, and Pseudonocardia, ferridoxin reductase and toluene monooxygenase participate in the degradation of xylene, and the metabolite is 3-methylbenzyl alcohol (Fig. 8C). The microorganisms associated with the degradation metabolism of ethylbenzene were pseudomonas, Bacillus, Actinomycetota, Bacteroidota and Euryarchaeota. Many oxygenases, such as ferredoxin NAD(P) reductase, naphthalene dioxygenase, arylnitro and dinitrotoluene, participate in ethylbenzene degradation, and the metabolite is 1-phenylethanol (Fig. 8D).

4. Discussion

This study provides a comprehensive metabolic characterization of cigar tobacco leaves from Yunnan and major foreign production regions. The PCA and HCA results (Fig. 3) revealed clear clustering and separation trends between Yunnan and Foreign CTLs, supporting the rationale for comparative analysis between these two groups. However, it is important to acknowledge the limitations of the sampling strategy. With only 25 samples representing nine geographically dispersed regions, the statistical power to account for within-region variability is limited. Additionally, factors such as inter-annual variation, differences in cultivars, post-harvest processing, and fermentation methods may contribute to the observed metabolic differences. Despite these constraints, the selected samples exhibited significant group-wise separation, justifying a preliminary exploration of origin-related metabolic signatures.

In terms of conventional chemical components, Foreign CTLs generally exhibited higher sugar, protein, and polyphenol contents, whereas Yunnan CTLs had significantly higher nicotine levels—a trend consistent with previous reports comparing domestic Chinese tobaccos with those from South America and Southeast Asia [12,13]. The elevated starch, pectin, and petroleum ether extract contents in Yunnan samples suggest incomplete hydrolysis of macromolecules or interruptions in metabolic pathways during curing and fermentation. This may explain the lower aromatic activity and differences in sensory quality previously noted in Yunnan CTLs [5,14]. Volatile aroma profiling further highlighted these differences. Six aroma components, including pyridine, megastigmatrienone, and β-damascenone, were more abundant in Yunnan CTLs, whereas 12 others, such as β-cyclocitral and 1-penten-3-one, were predominant in Foreign samples (Fig. 1). These findings align with earlier comparative studies of Cuban, non-Cuban, and Chinese cigars [14].

Currently, over 6000 chemical components have been identified and reported in tobacco [15], and the basic skeleton substances in tobacco leaves consist of sugars, nitrogen-containing compounds, alkaloids, pigments, and mineral elements, as well as a variety of secondary metabolites. In this study, through untargeted metabolomics, we identified phenolic acids, alkaloids, flavonoids, amino acids and derivatives, lipids, and organic acids as the major non-volatile components of CTLs (Fig. 2A), while volatile profiles were dominated by terpenoids, lipids, heterocyclic compounds, ketones, and hydrocarbons (Fig. 2B). The dynamic nature of these metabolites during air-curing and fermentation has been well documented [1,10,16], underscoring the importance of post-harvest processing in shaping the final metabolic landscape.

To identify potential biomarkers, we applied stringent criteria (AUC = 1.0, Log₂FC > 2, VIP >1, P < 0.01) and selected the top three candidates from each metabolite category. These include nicotine, reducing sugar, and pectin from conventional components; pyridine, megastigmatrienone, and β-damascenone from aroma compounds; tomatine, N-(3-indolyl)-l-alanine, and 5,6,7-trimethoxycoumarin from non-volatiles; and bornyl acetate, 5,6,7,8-tetrahydroquinoxaline, and benzoic acid 1-methylethyl ester from volatiles. Although these compounds show strong discriminative power in this dataset, they should be regarded as candidate biomarkers pending validation in independent, larger-scale studies. The multi-metabolite panel approach, as suggested by Dan et al. [17] and Steinfath et al. [18], may offer more robust origin discrimination than single compounds alone.

MetOrigin analysis provided further insight into the microbial and enzymatic underpinnings of metabolic variation. Non-volatile biomarkers were linked to pathways such as flavonoid degradation, linoleic acid metabolism, and biosynthesis of cutin, suberin, and wax, whereas volatile biomarkers participated in ethylbenzene and xylene degradation (Fig. 8). Microbial taxa including Pseudomonas, Bacillus, and Sphingomonas were implicated in these processes, consistent with prior studies highlighting the role of bacterial communities in CTL fermentation and flavor development [19]. These results suggest that geographic differences in CTL metabolism are influenced not only by soil and climate but also by microbial ecology and enzymatic activity during processing.

In conclusion, while this study identifies promising metabolic discriminants between Yunnan and Foreign CTLs, the findings are preliminary and require further validation. Future work should incorporate larger sample sizes, standardized processing conditions, and targeted validation of candidate biomarkers. Moreover, manipulating microbial consortia and enzymatic activities during fermentation may help modulate carbon and nitrogen metabolism, offering a pathway to improve the quality and homogeneity of Yunnan CTLs.

5. Conclusion

This multi-omics study systematically delineates the distinct metabolic signature of Yunnan CTLs in comparison to foreign counterparts. We demonstrate that Yunnan CTLs are characterized by a lower sugar-to-nicotine ratio and higher contents of starch, pectin, and lignin, a chemical profile that provides a plausible explanation for their documented sensory drawbacks. We identified 12 potential discriminatory metabolites, including nicotine, reducing sugar, pectin, pyridine, trilinolein, and tomatine, etc. These substances have high diagnostic value in differentiating Yunnan CTLs from those from abroad. The differential metabolites are implicated in key pathways like alkaloid biosynthesis and terpenoid backbone formation, and their profiles are linked to specific microbial communities (Pseudomonas, Bacillus, Sphingomonas) known to be active during fermentation. These findings suggest that the quality gap is not immutable but can be addressed by strategic interventions. Future efforts should focus on leveraging this knowledge to guide fermentation technology. Specifically, the targeted use of microbial consortia and enzyme preparations to modulate the carbon-nitrogen coupling metabolism presents a promising strategy that can degrade accumulated macromolecular substances and promote the conversion to desirable aroma compounds.

Author contributions

ZG (Guanghai Zhang): writing—original draft; methodology; validation; investigation; data curation. ZG (Gaokun Zhao) and ZG (Guanghai Zhang): funding acquisition; resources; writing—review & editing; supervision. LM, WY, YH, LW and XH: validation, investigation, discussed the results and commented on the manuscript. All authors have read and agreed to the published version of the manuscript.

Declaration of competing interest

There are no conflicts of interest in the submission of this manuscript, and manuscript is approved by all authors for publication.

Appendix A. Supplementary data

The following is/are the supplementary data to this article:

Data availability

Data will be made available on request.