One milliliter of preserved Staphylococcus saprophyticus culture was inoculated into liquid LB medium and incubated at 200 rpm for 12 h to prepare the seed culture. This seed culture was subsequently transferred into a fermenter under the following conditions: natural pH, aeration rate of 1 m³/h, and agitation at 400 rpm. After 12 h, the culture was centrifuged, and the resulting pellet was washed three times with distilled water and stored at 4 °C for further use.

In the control group (CK), only water was added, while in the experimental group (J), Staphylococcus saprophyticus was added at 1.3% of cigar filler leaves weight. Both groups were re-moistened to moisture content of 34% ± 2%. Samples were collected in triplicate at four time points: before fermentation (CK-0, J-0), after the first overturn (CK-1, J-1; day 7), after the second overturn (CK-2, J-2; day 14), and after the third overturn (CK-3, J-3; day 35). All subsequent analyses were conducted using these biological replicate samples. The entire fermentation process lasted 35 days under controlled conditions (28 °C ± 2 °C, relative humidity 80% ± 2%). After fermentation, the leaves were dried at 75 °C, ground, and passed through a 60-mesh sieve.

Ten grams of sieved sample were placed in a 1,000 mL round-bottom flask, to which 200 mL of saturated saline solution was added. In parallel, 60 mL of dichloromethane was added to a 100 mL round-bottom flask. Both flasks were connected to the Simultaneous Distillation Extraction (SDE) apparatus. The dichloromethane phase was heated in a 50 °C water bath, and the aqueous phase was heated in a 140 °C oil bath. Once a stable liquid-liquid interface was formed, extraction was carried out for 5 h. The organic phase was then collected and supplemented with 50 μL of 1.2028 mg/mL ethyl benzoate as the internal standard. The extract was concentrated to 2 mL using a rotary evaporator, dried over anhydrous sodium sulfate, filtered through a 0.22 μm organic phase membrane, and transferred to GC-MS vials for analysis.

Neutral aroma compounds were analyzed using an Agilent 7890A gas chromatograph equipped with an HP-5MS capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness). Samples (1 μL) were injected in split mode with a 10:1 split ratio. Helium was used as the carrier gas at a flow rate of 1 mL/min. The temperature program was as follows: initial temperature of 40 °C held for 2 min, ramped at 2 °C/min to 200°C (held for 5 min), and then increased at 10 °C/min to 280 °C. Detection was performed using an Agilent 5975C mass spectrometer with a transfer line temperature of 250 °C and an ion source temperature of 230 °C under 70 eV electron impact ionization.

Chromatographic peaks obtained from GC-MS were matched to reference spectra using the NIST 14.0 mass spectral library (Agilent Technologies). The compound with the highest match was selected for identification. Quantification was performed using phenylethyl acetate as the internal standard. The concentration of each volatile compound was calculated using the formula: Wi = (30 × Mi)/(Si × 2/10), where Wi is the content of each volatile aroma compound (μg/g), Mi is the peak area of the compound, and Si is the peak area of the internal standard.

A portion of fermented tobacco was hand-rolled into cigars (14 mm in diameter, 110 mm in length). Sensory evaluation was conducted by trained experts from Enshi Cigar Tobacco Production Base using a nine-point scale (see Supplementary Table S1). The evaluation included eight indicators: aroma quality, aroma intensity, aftertaste, off-notes, sweetness, irritation, combustibility, and ash color. Each sample was evaluated in triplicate.

Processed tobacco samples were immersed in 80 mL of sterile distilled water at room temperature for 15 min. The suspension was subjected to 10 s of centrifugation, followed by 90 s of vortex mixing. The mixture was centrifuged at 500 × g for 1 min to remove leaf debris, then at 10,000 × g for 10 min to collect microbial cells. The pellet was washed with a buffer containing 100 mmol/L Tris-HCl, 50 mmol/L EDTA-Na₂, 20 g/L PVP, 1 mL/L Tween-20, and 1.4 mol/L NaCl (pH 8.0). The mixture was rotated for 45 s at room temperature, incubated in a 65 °C water bath for 5 min, and centrifuged at 6,000 × g for 6 min. The washing steps were repeated until the supernatant became nearly colorless. The resulting microbial pellet (≥200 mg) was snap-frozen in liquid nitrogen for at least 20 min and stored at −80 °C.

DNA extraction was performed using the CTAB method. Briefly, 1,000 μL of CTAB lysis buffer, lysozyme, and an appropriate amount of sample were added to a 2.0 mL EP tube and incubated in a 65 °C water bath with occasional inversion. The lysate was centrifuged, and the supernatant was extracted with phenol (pH 8.0):chloroform:isoamyl alcohol (25:24:1) and then with chloroform:isoamyl alcohol (24:1). After centrifugation, the supernatant was mixed with isopropanol and precipitated at −20 °C. The pellet was washed twice with 75% ethanol, air-dried, and dissolved in ddH₂O. RNase A (1 μL) was added to remove RNA, followed by incubation at 37 °C for 15 min.

Library preparation and sequencing were performed by Beijing Novogene Co., Ltd. DNA samples were fragmented, end-polished, A-tailed, and ligated with Illumina adapters, followed by size selection. PCR amplification (except in PCR-free protocols) and purification were performed using the AMPure XP system. Libraries were quality-checked using the Agilent Fragment Analyzer and quantified using Qubit and qPCR. Qualified libraries were pooled and sequenced on Illumina platforms.

Differences in microbial taxa before and after fermentation were analyzed using the cloud-based platforms provided by Beijing Novogene (https://magic.novogene.com/) and CNSknowall (https://cnsknowall.com). Raw sequencing data were processed using Fastp to obtain clean reads. Functional and taxonomic annotation was conducted using DIAMOND. Statistical significance was assessed using Statistica 23.0 software (SPSS (Statistical Package for the Social Sciences) Inc., Chicago, IL, United States of America, https://spss.updatestar.com).

The distribution of the microbial community composition at the OTU level on the surface of cigar filler tobacco leaves is shown in Figure 2. Before fermentation (CK-0), the surface OTU count reached its maximum at 2,374. After three overturns, the number of OTUs declined to 2,283, 1,965, and 2,152, respectively for natural fermentation. In contrast, the inoculated fermentation group (J-1, J-2, J-3) exhibited a more pronounced decline, with OTU counts decreasing to 1,667, 1,425, and 1,461, respectively.

The results of α-diversity indices in Figure 3 showed that, by the third overturn, both Chao1 and Shannon indices were significantly reduced in the Staphylococcus saprophyticus-inoculated group, indicating a decline in microbial richness and diversity. This reduction may be attributed to the ecological dominance of S. saprophyticus, which likely outcompeted and inhibited the growth of other microbial taxa (Sosa-Fajardo et al., 2024). Similar observations have been reported in previous studies, where the inoculation of microorganisms such as Acinetobacter sp. 1H8 and A. indicus 3B2 led to decreased Chao1 and Shannon indices (Zheng et al., 2022), and the addition of Bacillus velezensis was also associated with a reduction in Shannon indices compared to natural fermentation (Zhang, 2023).

Moreover, a reduction in the number of bacterial OTUs in the natural fermentation group relative to the pre-fermentation stage was found, particularly after the second overturn (p < 0.001) and third overturn (p < 0.05). This finding contrasts with earlier reports, which documented an increase in OTU abundance during fermentation (Yao et al., 2024). This discrepancy may be driven by the combined effects of several ecological mechanisms, including resource depletion, microbial antagonism, and environmental selection pressure. During fermentation, dominant taxa may rapidly consume the limited nutrients available in the system, thereby constraining the growth and expansion of other microbial populations (Nwogwu, 2022). Concurrently, certain strains may secrete antimicrobial metabolites—such as organic acids, hydrogen peroxide, or bacteriocins—that further inhibit the proliferation of potential competitors. Microorganisms with strong adaptability to evolving fermentation conditions are more likely to persist through natural selection, whereas less adaptable strains are gradually eliminated from the community (Sudiartha et al., 2023; Xing et al., 2021). Moreover, under dual pressures of nutrient scarcity and environmental stress, genetic drift may also occur within the microbial population, resulting in a stochastic decline in overall community diversity (Wani et al., 2022; Lynch et al., 2016).

Mechanistically, the early metabolic activity of Staphylococcus saprophyticus may have accelerated the depletion of available nutrients in the system while promoting the accumulation of fermentation-derived byproducts such as organic acids, alcohols, and reactive oxygen species. These metabolites can significantly alter the physicochemical properties of the fermentation environment, for instance by reducing pH, increasing osmotic pressure, and disrupting redox balance, thereby suppressing the growth and metabolic activity of microorganisms with lower environmental adaptability (Guan and Liu, 2020). This ecological filtering effect favors the survival of nutrient-tolerant strains and ultimately contributes to a contraction in overall community diversity.

As fermentation progressed, particularly during the third overturn, environmental changes such as enhanced oxygen diffusion may have supported the partial recovery of facultative aerobic or subdominant microbial taxa. Increased oxygen availability can alleviate anaerobic stress and facilitate more efficient energy production via oxidative phosphorylation, thus improving microbial competitiveness under nutrient-limited conditions (Li et al., 2020; Yang et al., 2024). Nevertheless, despite this partial rebound, the sustained ecological dominance of S. saprophyticus—through continued resource competition, production of inhibitory metabolites, and stable niche occupation—likely constrained the re-establishment of a diverse microbial community, resulting in persistently lower microbial diversity compared to the natural fermentation group.

The growth and metabolic activity of microbial communities are critical determinants of cigar tobacco leaves (Huang et al., 2021). The microbial community of cigar filler leaves was primarily composed of four phyla (Figure 4a): Bacillota, Pseudomonadota, Ascomycota, and Basidiomycota.

At the genus level, the microbial community of cigar filler leaves was dominated by four genera (Figure 4b): Bacillus, Staphylococcus, Aspergillus, and Sphingomonas. In the non-fermented group, Sphingomonas exhibited the highest relative abundance, whereas Staphylococcus and Bacillus were less prevalent. During natural fermentation, Sphingomonas showed a decreasing trend initially, followed by a resurgence, while Aspergillus increased significantly, peaking toward the end of fermentation. The relative abundances of Staphylococcus and Bacillus gradually increased over time.

In the S. saprophyticus-inoculated group, Bacillus showed extensive proliferation, with its relative abundance exceeding 92% at the end of fermentation and peaking at all three turnover points. Staphylococcus reached its maximum abundance after the first overturn and gradually declined thereafter, a trend mirrored by Sphingomonas and Aspergillus. These shifts indicate that the introduction of S. saprophyticus significantly reshaped the microbial ecosystem on the surface of the cigar filler leaves, accelerating microbial succession and altering both the composition and diversity of the microbial community. LEfSe analysis results were shown in Supplementary Figure S3.

As one of the key bacterial taxa in tobacco fermentation, Bacillus has been shown in numerous studies to play a pivotal role in enhancing tobacco quality. For example, Bacillus cereus has been reported to degrade hemicellulose and cellulose, improving the physical properties of wrapper leaves (LI, 2022). Dominant bacteria in cigar leaves are also capable of degrading aromatic hydrocarbons, aliphatic compounds, and lignin (Liu et al., 2021). Moreover, Bacillus species can hydrolyze tobacco proteins, thereby reducing bitterness and astringency (Chen Xing, 2015). Other species such as Rhodococcus, Bacillus, and Pseudomonas are effective in degrading nicotine (Zhong et al., 2010; Liu et al., 2015). Additionally, Bacillus has been found to break down starch (Pen et al., 1992), cellulose (Li et al., 2006), and other macromolecular substances, thereby reducing harshness. Collectively, these findings highlight the functional versatility of Bacillus in decomposing complex compounds in tobacco leaves, which improves fermentation efficiency and enhancing product quality.

According to KEGG metabolic pathway analysis, the five most abundant functional categories were carbohydrate metabolism, amino acid metabolism, membrane transport, energy metabolism, and metabolism of cofactors and vitamins, consistent with previous findings (Yao et al., 2024). For further details, refer to Supplementary Figure S4. Functional gene annotation results conducted using the EggNOG database was shown in Figure 5.

Notably, the significant enrichment of genes involved in nucleotide transport and metabolism suggests enhanced nucleic acid turnover during fermentation, including key processes such as DNA replication, RNA transcription, repair, and degradation (Lane and Fan, 2015). In metabolically active fermentation environments, microbial communities necessitate rapid genome replication to adapt to environmental stresses and sustain ecological competitiveness. Furthermore, accelerated synthesis of rRNA and tRNA likely enhances translational activity, thereby promoting the initiation and progression of various downstream secondary metabolic pathways. Consistently, the enrichment of genes related to transcription, translation, and ribosomal structure indicates a significant increase in microbial growth rates and protein biosynthesis during fermentation (Zang et al., 2024; Zhang et al., 2022). This indicates that during specific fermentation stages, the efficient synthesis of metabolic enzymes, transporters, and transcriptional regulators represents a core biological mechanism underpinning system-wide metabolic activity and transformation efficiency.

Moreover, the persistently high abundance of genes associated with carbohydrate and amino acid metabolism highlights the central importance of these pathways during fermentation, aligning with KEGG Level 2 functional categorization. Enhanced carbohydrate metabolism not only facilitates rapid energy production through glycolysis and the pentose phosphate pathway (PPP), but also supplies precursor substrates for the biosynthesis of lipids, alkaloids, and other secondary metabolites. Simultaneously, activated amino acid metabolism supports protein synthesis and regulates aroma precursor formation and microbial signaling, thereby playing a pivotal role in modulating fermentation dynamics (Guo et al., 2023).

To further explore functional contributions, the relationship between enriched microbial genes and EggNOG functional categories was analyzed in the S. saprophyticus-inoculated group (Supplementary Figure S5). Sankey diagram results showed that Aspergillus was associated with all 24 functional categories, followed by Sphingomonas (22), Pseudomonas (21), Staphylococcus (20), and Bacillus (19). These associations indicate that the introduction of S. saprophyticus not only enhanced gene enrichment in these genera but also significantly increased the abundance of genes in corresponding metabolic pathways. This functional enhancement likely improved metabolic efficiency, stimulated microbial activity, and influenced the final outcomes of fermentation and flavor formation. Functional gene annotation results based on the CAZy database are provided in the Supplementary Figure S6.

Overall, the persistently high levels of the glycosyltransferase, carbohydrate-binding module, and carbohydrate esterase genes in the experimental group likely facilitated the degradation and transformation of complex sugars during fermentation, significantly shaping the chemical composition and flavor attributes of the cigar filler leaves. These results align with previous findings demonstrating that dominant metabolic enzymes—such as glycoside hydrolases (GHs) and glycosyltransferases (GTs)—play critical roles in the natural fermentation of cigar tobacco (Wu et al., 2023).

Microbial interactions are pivotal in shaping the community structure during fermentation processes (Mould and Hogan, 2021). As illustrated in Figure 6, a strong positive correlation was observed between Staphylococcus and Bacillus, both of which showed significant negative correlations with most of the other identified genera. This synergy may indicate cooperative nutrient utilization or metabolic exchange, facilitating mutual growth, while their antagonistic relationships with other taxa may stem from competition over resources or alteration of the local environment.

Bacillus and Staphylococcus saprophyticus are particularly noteworthy for their functional roles in the fermentation of cigar tobacco leaves. Both genera secrete a diverse array of extracellular enzymes—such as proteases, amylases, lipases, xylanases, and cellulases—that synergistically degrade macromolecular compounds including proteins, starches, lipids, and plant cell wall components. This enzymatic activity releases low-molecular-weight substrates that support microbial growth and promote tobacco maturation (Kumar et al., 2020; Yin et al., 2025). Although these two taxa may compete for initial abundant substrates (e.g., carbohydrates and proteins), their complementary enzymatic profiles facilitate more efficient decomposition of a wider range of compounds. For example, S. saprophyticus-derived xylanases and cellulases break down plant structural polymers, exposing additional carbon sources that benefit subsequent degraders like Bacillus (Si et al., 2023), thereby accelerating the overall fermentation process.

Acinetobacter exhibited a negative correlation with Aquabacterium but a positive correlation with Bacillus, suggesting both competitive and cooperative dynamics. Such patterns align with previous reports indicating strain-specific interactions—for instance, Acinetobacter indicus 3B2 inhibits Aquabacterium while promoting Bacillus, thereby influencing broader community dynamics involving Staphylococcus and Aerococcus. These interdependent relationships collectively drive microbial succession and impact fermentation outcomes.

Using Spearman correlation analysis was performed the relationships between 64 volatile compounds and 20 representative bacterial genera were analyzed. As shown in Figure 7, Staphylococcus exhibited positive correlations with 37 aromatic compounds, including 1-methyl-4-(2-methyloxiranyl)-7-oxabicyclo [4.1.0] heptane, cis-6-nonenal, and (Z)-7-hexadecenal, while showing negative correlations with 21 compounds such as 2-methylpyrazine, benzaldehyde, and isopulegol. Bacillus was positively correlated with 40 aroma compounds—including solanone, phenylethyl alcohol, damascone, and 6-methyl-5-hepten-2-one—and negatively correlated with 20 compounds, such as pyrrole, furfuryl alcohol, and (E)-4-nonenal. These findings highlight the central role of microbial metabolism in shaping the volatile flavor profile of cigar filler leaves.

Bacillus is a ubiquitous genus in natural environments, particularly in tobacco, where it plays a key role in fermentation by degrading macromolecules such as carotenoids into low-molecular-weight aromatic compounds. Previous studies have demonstrated the ability of Bacillus to enhance both the aroma and overall quality of tobacco products (English et al., 1967). For instance, fermentation with B. velezensis was reported to increase total aroma compounds in cigar leaves by 26.1%. Likewise, co-fermentation of Bacillus megaterium with yeast strains isolated from tobacco leaves led to a 56.4% increase in total volatile compounds, including elevated levels of solanone, further validating their contribution to aroma enhancement (Mao, 2023). In the present study, the solanone content in the inoculated group increased significantly to 67.4 μg/g, representing 170.7% increase.

Although studies on Staphylococcus in cigar fermentation remain limited, its role in meat fermentation provides valuable insights. In meat products, Staphylococcus species modulate microbial and biochemical dynamics, particularly processes involving proteins and fats, thereby influencing product quality and enhancing the formation of volatile flavor compounds (Li et al., 2015). For example, Staphylococcus xylosus has been shown to significantly increase the production of alcohols and esters in sausages, enriching the volatile flavor profile. Similarly, inoculation with S. epidermidis, S. xylosus, and S. auricularis into dry-cured meat led to the generation of distinct aroma compounds: S. epidermidis produced elevated levels of heterocyclic compounds such as 2,4,6-trimethylpyrimidine; S. xylosus enhanced the formation of aldehydes including 2-methylbutanal and 3-methylbutanal; and S. auricularis promoted the synthesis of esters like 6-deoxymannonic lactone.

In this study, the volatile compounds identified in cigar filler leaves were predominantly aldehydes and alcohols, which differed from those typically found in meat fermentations, likely due to the distinct substrate compositions. Meat is rich in proteins, fats, and free amino acids, whereas tobacco leaves primarily contain cellulose, polyphenols, carbohydrates, and alkaloids such as nicotine. These substrate differences lead to divergent microbial metabolic pathways, resulting in the production of distinct volatile compounds during fermentation. Despite these differences, the observed contribution of Staphylococcus to aroma formation in cigar fermentation highlights its potential for improving sensory quality, although further research is needed to elucidate its specific metabolic mechanisms.