Introduction

The perception of bitterness taste is an important defense mechanism for human body to prevent the harm of toxic substances1. Various bitterness substances, including alkaloids, terpenoids, peptides, glycosides, salts, and amino acids, have been identified from a wide range of sources such as food plants, medicinal plants, animal foods, and food processing2. The taste mechanism of bitter substances is quite complex. Human bitterness taste receptors (hTAS2Rs) are chemosensory receptors that belong to the G protein-coupled receptor (GPCR) family3. A total of 25 hTAS2Rs have been identified and reported, which can bond with structurally diverse compounds related to bitterness taste. As an important computer-aided simulation method, molecular docking technology provides an effective tool for the depth analysis of the interaction mechanism between bitterness components and receptors in recent years4. Moreover, the cryogenic electron microscopy structures of TAS2R46 and TAS2R14 have been identified, which provided insights into the ligand recognition and interaction mechanism of bitterness taste receptors1,5,6. Through molecular docking and transcriptomics analysis, the bitterness peptides of Jinhua ham and its taste mechanism were revealed7. The key bitterness components such as epigallocatechin gallate and epicatechin gallate of Wuyi rock tea were revealed through sensory evaluation and cell-based bitterness taste responses8. During Maillard reaction, several amino-carbonyl compounds that associated with bitterness taste were formed and influenced the quality of foods3.

Cigars, a special kind of tobacco products, are highly favored by consumers due to their rich aroma, full taste, and strong satisfaction9. Sometimes, cigar tobacco leaves (CTLs) possess the characteristics of high irritation and bitterness taste, and several cigar brands are known for their special bitterness taste. Moderate bitterness can enhance the complexity and richness of cigar smoke, while excessive bitterness could highly bring sensory discomfort and limit consumers’ sensory experience10. Although the bitterness components of various plant-based foods and drugs have been revealed, the key compounds associated with bitterness taste of cigar tobacco leaves are not known clearly10,11. Molecular docking technology is expected to provide a new research perspective for revealing the key bitterness compounds of CTLs. In this study, three bitterness CTLs along with one reference CTL were chosen (Fig. 1). The sensory evaluation was carried out, and the volatile components of mainstream smoke particulates were identified by gas chromatography-mass spectrometry (GC-MS). The key potential bitterness compounds were screened out using orthogonal partial least squares discriminant analysis (OPLS-DA), partial least squares regression (PLSR), and correlation heatmap analysis. Subsequently, the key bitterness compounds were verified through sensory evaluation, and the interaction mechanism of the key bitterness compounds with bitter receptors were determined through molecular docking and kinetic simulation.

Fig. 1
Fig. 1
Full size image

The research approach of this article.

Materials and methods

Materials and instruments

Test tobacco leaves were chosen as F447-1, F166-1, and B065-2 tobacco leaves (Indonesia CTLs). Chuxue 14 tobacco leaves (Hubei province, China) without bitterness taste were selected as reference tobacco leaves.

Myosmine, cotinine, nicotinamide, 3-ethyl-pyridine, 2,3’-bipyridine, and phenyl acetate were purchased from Shanghai Macklin Biochemical Technology Co., Ltd. (Shanghai, China). Nicotine was obtained from Zhengzhou Tobacco Research Institute of ZNTC.

GC-MS was determined on Agilent 6890/5973 GC-MS (Agilent, USA). Mainstream smoke particulates were collected using a RM20H rotary smoking machine (Bowat Casey, Germany), which were extracted using a ZHWY-304 shaker (Shanghai Zhicheng Instrument Manufacturing Co., Ltd., China).

Sensory evaluation method

CTLs were maintained at a constant temperature of 22 ± 1 °C and a relative humidity of 60 ± 3% to balance the water content. Subsequently, the tobacco leaves were rolled into experimental cigars with a diameter of 1.0 cm and a circumference of about 3.1 cm. Cigar samples with a weight of 2.0 ± 0.1 g were chosen for sensory evaluation.

The sensory evaluation team was composed of 12 experienced evaluators with a minimum of 5 years of experience in tobacco product assessment. The sensory evaluation was performed according to the Chinese standard YC/T 497–2014, and the main sensory evaluation aspects were selected as bitterness, sweetness, aroma quality, smoothness, oral irritation, oral residue, astringency, throat irritation, throat dryness, nasal irritation, and offensive odor.

All samples were coded with random three-digit numbers and presented in a randomized order to each panelist to eliminate order effects. The sensory evaluation was performed by the evaluators under blinded conditions, with no access to the identities of the samples. During a single evaluation, each evaluator sequentially assessed all four samples, with a mandatory 2-minute rest interval between samples. During this interval, evaluators rinsed their mouths with purified water to mitigate sensory fatigue and cross-over effects. Each product was assessed by a trained panel in three independent sessions. The average scores were calculated.

The sensory evaluation was conducted using a 10-point scale system (scores 0–10). For aspects including bitterness, sweetness, aroma quality, and smoothness, higher scores indicated stronger characteristic intensity. For the other negative aspects such as oral irritation, oral residue, astringency, throat irritation, throat dryness, nasal irritation, and offensive odor, higher scores represented weaker negative characteristic intensity.

All the methods were carried out in accordance with relevant guidelines and regulations. All the experimental protocols including sensory evaluation were approved by the Institutional Review Board of Zhengzhou University of Light Industry (IRB No. 2025-0008-12). All the evaluators agreed the disclosure and usage of the sensory evaluation data, and approved the final version of the manuscript. An informed consent was obtained from all the evaluators.

Collection of mainstream smoke compounds

The stems of four CTLs were removed and the water content was maintained to 18–21%. Subsequently, the leaves were made into cut tobaccos and rolled into cigarettes (0.8 ± 0.05 g). Each of 40 test cigarettes was selected to collect mainstream smoke compounds according to ISO 3308 standard using an RM20H rotary smoking machine. The suction volume was 35 mL with a duration of 2.0 s, and the suction interval was 60 s. A Cambridge filter paper with diameter of 92 mm was used to collect the mainstream smoke compounds of 40 cigarettes. After suction, the Cambridge filter papers were placed in conical flasks, and 50 mL of dichloromethane was added. The ultrasonic extraction was performed for 30 min, and followed by a shaken extraction for 40 min. Then 10 mL of the CH2Cl2 solution was concentrated to 1 mL in a water bath. The concentrated solution was filtered by organic membrane and 100 µL of phenyl acetate (1 mg/mL) was added as internal standard, which was further analyzed by GC-MS.

GC-MS analysis

GC-MS was performed on an Agilent 6890 − 5973 gas chromatograph-mass spectrometer that equipped with an DB-5 MS column (60 m × 0.25 mm i.d. × 0.25 μm d.f.) with an injection volume of 1.0 µL. The carrier gas was high-purity helium and the flow rate was 1.0 mL/min with a split ratio of 5:1. The initial temperature was 50 °C, which increased from 50 °C to 250 °C at a rate of 2 °C/min and maintained for 10 min. The temperature of transmission line was 280 °C, and ion source temperature was 230 °C. The ionization mode was EI (electron energy, 70 eV), and the mass scanning range was set to 35–550 amu.

Qualitative and quantitative analysis

For qualitative analysis, the GC-MS data (Supporting Information Figure S1 to S3) were analyzed in the NIST20 standard mass spectrometry, and the compounds with a matching scores of over 85 were selected. The key volatile compounds were further qualitatively verified by their GC-MS retention time with those of the standard compound kept in the author’s laboratory11. The retention index (RI) for each compound was calculated and verified by comparison with reference values from the NIST Chemistry WebBook and published literature11. The RI was determined using the formula:

$$RI = 100 \times {\text{n}} + \frac{{t_{a} - t_{n} }}{{t_{{{\text{n}} + {\text{1}}}} {\text{ - t}}_{n} }} \times 100$$

In the formula, n is the number of carbon atoms (Cn) of a series of alkanes. ta means the retention time of compound a (the retention time of compound a is between Cn and Cn + 1). tn and tn + 1 are the retention time of n-alkanes with n and n + 1 carbon atoms, respectively.

For quantitative analysis, the internal standard method was preliminarily employed using phenyl acetate as the internal standard. Specifically, each of 100 µL of phenyl acetate solution (1 mg/mL) was added to the selected samples. A specific calculation formula was as follows: the content of target compound = (peak area of target compound/peak area of internal standard) × content of internal standard.

For key bitterness compounds, GC-MS standard curves were established. A stock solution of the internal standard phenethyl acetate (1.0 mg/mL) was prepared. Standard bitterness compounds were accurately weighed, dissolved in ethanol, and transferred into volumetric flasks. The bitterness compound mixture solutions were gradually diluted to obtain serial working solutions with concentrations of 1, 5, 10, 25, 50, 75, and 100 µg/mL with 125 µL of internal standard solution. GC-MS analysis was conducted on the mixed standard working solution. Standard curves were established using the corresponding chromatographic peak area ratio (y) based on the mass concentration ratio (x) of each bitterness compound with the internal standard. The standard curves, linearity ranges, correlation coefficients, relative standard deviation, LOD, and LQD of key bitterness compounds were provided in the Supporting Information Table S1.

Screening principles of key bitterness compounds

The volatile bitterness compounds were screened out according to their potential contribution to smoke bitterness, using tobacco smoke components listed in the book “The chemical components of tobacco and tobacco smoke” by Rodgman and Perfetti as a primary data sources12. The selection was performed according to the following four rules. First, the non-terpene aliphatic alkanes were removed, since their taste activity and contributions were weak, such as hexacosane and triacontane. Second, the saturated long-chain fatty acids, alcohols, and esters were deleted due to their weak taste contribution, such as tetradecanoic acid and n-hexadecanoic acid. Third, the compounds obviously related to sweetness taste were removed such as 2-methyl-2-cyclopenten-1-one, 3-methyl-2-cyclopenten-1-one, and 2,3-dihydro-3,5-dihydroxy-6-methyl-4H-pyran-4-one (DDMP). Last, compounds highly associated with special aromas such as styrene, 1-phenylethyl acetate, 2-butyl-3-methylpyrazine, and alloaromadendrene were also excluded, since they were hardly related to bitterness in combination with literatures13,14. The excluded compounds with corresponding justification were provided in the Supporting Information Table S2.

Smoke sensory evaluation verification of key bitterness compounds

According to the Chinese standard YC/T 497–2014, the selected key compounds associated with bitterness taste were verified by smoke sensory verification11,15. The selected compounds including nicotine, myosmine, 2,3’-bipyridine, nicotinamide, 3-ethyl-pyridine, and cotinine were prepared into aqueous solution with a concentration of 1000 µg/mL, Then 20 µL of each solution was injected into the test cigar made by Chuxue 14 tobacco leaves via central injection methodology. The control group was injected with the same volume of purified water. The sensory evaluation team was composed of 12 evaluators, and the average scores of bitterness taste were calculated using the scores given by the evaluators.

According to GB/T 12312 − 2012, the bitterness taste of the evaluators was calibrated and trained using the quinine hydrochloride standard solution to improve the accuracy and reproducibility of the evaluation results. As shown in Table 1, the bitterness taste is divided into five levels, and each level is given a certain range of bitterness score and standard16.

Table 1 Standard grade and scores for bitterness taste.
Full size table

Taste verification of key bitterness compounds

The selected key bitterness compounds (nicotine, myosmine, 2,3’-bipyridine, and nicotinamide) were prepared in aqueous solutions with different concentrations (1–300 µg/mL). The evaluation group was composed of 12 experts, which were trained using the above-mentioned quinine hydrochloride standard solution. The taste evaluation was carried out according to the above-mentioned method, and the average scores were calculated.

Molecular docking of compounds with bitter receptors

The protein structures of TAS2R14 and TAS2R46 were obtained from RCSB PDB database (https://www.rcsb.org/). TAS2R1, TAS2R7, TAS2R8, TAS2R9, TAS2R10, TAS2R13, and TAS2R16 were downloaded from AlphaFold database (https://alphafold.com/). The three-dimensional structures of the key bitterness compounds were obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). Auto Dock 4.2.6 was used for molecular docking, and the hydrogenation and charge optimization of proteins (receptors) and bitterness compounds (ligands) were also carried through this software17.

The binding site of the protein was determined based on the docking region reported in the literature using Discovery Studio 2021 Client18. According to the dimensions of the identified protein active site, a grid box was placed within the protein to identify ligand interaction or docking binding sites19. The grid box dimensions were set to 60 × 60 × 60, with its center positioned at X = 138.8, Y = 137.3, and Z = 154.9. After setting the grid size, molecular docking was performed using AutoDock Vina. All the ligands were docked, and nine conformations and corresponding docking scores for each ligand were generated. In combination with BIOVIA Discovery Studio 2016 Client, 3D and 2D visual analysis of the binding process was performed to provide an intuitive structural interpretation of molecular interactions20.

Molecular dynamics simulation

Molecular dynamics simulation was performed using YASARA program. The program core, docking method, and parameter settings have been integrated in the YASARA programs, which led to the easy handle for molecular dynamics simulation. The docking protein-ligand complex was placed in a 90 Å × 60 Å × 65 Å cube box, which was set as a periodic boundary condition20. And the system contained an appropriate amount of Na+ or Cl− to simulate the physiological environment. The environment of molecular dynamics simulation was set to 25 °C with pH of 7.4. The data were collected every 100 ps, and the molecular dynamics simulation of the complex system was performed for 50 ns. The root mean square deviation (RMSD) and root mean square fluctuation (RMSF) of the system were analyzed.

Data analysis

The radar map and correlation heatmap were drawn by Origin 2018 software. OPLS-DA was implemented through SIMCA 14.1. PLSR analysis was carried out using Unscrambler software. For molecular docking studies, AutoDock 4.2.6 was employed, and the subsequent visualization and analysis of docking results was performed in Discovery Studio 2016 Client. All experiments were performed in triplicate. The results are presented as the mean ± standard deviation (SD) from three independent measurements.

Results and discussion

Sensory evaluation results of CTLs

The sensory evaluation of 4 CTLs was carried out and the results were shown in Table 2; Fig. 2. The main sensory evaluation aspects were selected as bitterness, sweetness, aroma quality, smoothness, oral irritation, oral residue, astringency, throat irritation, throat dryness, nasal irritation, and offensive odor. For aspects including bitterness, sweetness, aroma quality, and smoothness, higher scores indicated stronger characteristic intensity. For the other negative aspects such as oral irritation, oral residue, astringency, throat irritation, throat dryness, nasal irritation, and offensive odor, higher scores represented weaker negative characteristic intensity.

Table 2 Sensory evaluation results of 4 CTLs (scores 0–10).
Full size table

It was interesting to note that bitterness taste (score 9.13 ± 0.03) of F447-1 was stronger than the other samples, and it possessed a special bitterness taste similar to some herbal medicine with long time of bitterness aftertaste. The bitterness taste of F166-1 (score 7.01 ± 0.04) was obvious, while the astringency, oral irritation, and throat dryness were also strong. The bitterness taste intensity of B065-2 was moderate (score 5.52 ± 0.05), and its offensive odor and irritation were relatively weak. The bitterness taste of reference CTLs Chuxue 14 was weakest (score 1.49 ± 0.05) and its sweetness taste was the highest. The order of bitterness taste of 4 CTLs was revealed as F447-1 > F166-1 > B065-2 > Chuxue 14.

Fig. 2
Fig. 2
Full size image

Sensory evaluation radar map of 4 CTLs.

GC-MS analysis of mainstream smoke compounds of CTLs

As shown in Table 3, a total of 33 potential bitterness compounds in the mainstream smoke of 4 CTLs were preliminarily screened out, which could be classified as tobacco alkaloids (6 kinds), pyridines (6 kinds), heterocyclics (4 kinds), ketones (2 kinds), phenols (5 kinds), alcohols (3 kinds), acids (2 kinds), and others (5 kinds)11. Careful analysis of the contents of volatile compounds indicated that the contents of myosmine (2), 2,3’-bipyridine (4), cotinine (5), DL-nornicotine (6), 3-ethyl-pyridine (10), phenol (19), p-cresol (21), phytol (25), and nicotinamide (27) in F166-1, F447-1, and B065-2 CTLs were overall higher than those in Chuxue 14 with significant differences (P < 0.05). These 9 compounds may bring key contributions to the stronger bitterness taste of F166-1, F447-1, and B065-2 CTLs than that of Chuxue 14. Furthermore, the contents of myosmine (19.25 ± 0.97 µg/g), 2,3’-bipyridine (24.33 ± 0.8 µg/g), and cotinine (22.74 ± 0.31 µg/g) in the mainstream smoke of F447-1 were obviously higher than the other CTLs, which could be one of the key factors for the most intense bitterness taste of F447-1. However, the kinds and contributions of the mainstream smoke compounds were quite complicated, and the further screening and analysis were performed through OPLS-DA, correlation heatmap, and PLSR.

Table 3 GC-MS analysis of mainstream smoke compounds of 4 CTLs (µg/g).
Full size table

OPLS-DA analysis of mainstream smoke compounds of CTLs

The OPLS-DA model was established using SIMCA 14.1 software, and the results were shown in Fig. 3. All 4 CTLs were revealed within the 95% confidence interval with clear distribution characteristics. In particular, F166-1 was located in the first quadrant, while F447-1 was presented in the second quadrant. B065-2 and Chuxue 14 were assigned in the third quadrant and fourth quadrant respectively. The established OPLS-DA model was highly significant with a predictive index Q2 = 0.983, and the explanatory indices were determined as independent variables R2X = 0.979 and dependent variables R2Y = 0.994. These results indicated the interpretability and predictive capability of the model was fine13. As shown in Fig. 3B, R2 was determined as 0.321 after 200 permutation tests. The regression line of model Q2 (Q2 = −1.07) crossed with the abscissa and intersected with the negative half of the vertical axis, which suggested that the model was reliable and not over-fitting21.

Fig. 3
Fig. 3
Full size image

OPLS-DA model (A), validation model (B) and VIP score plot (C) of mainstream smoke compounds of 4 CTLs.

The variable importance in projection (VIP) score is crucial for compound screening in OPLS-DA. Higher VIP values indicate greater contribution of the compound to the overall sample differentiation22. Using the standard of VIP > 1 and P < 0.05, a total of 7 key compounds for differentiation (Fig. 3C; Table 4) were screening out, which were revealed as nicotine (1), myosmine (2), 2,3’-bipyridine (4), cotinine (5), 3-ethylpyridine (10), 4-pyridone (12), and nicotinamide (27). The VIP of phytol (25) was determined as 1.646 but P value was disclosed as 0.3323, thus this compound was excluded. The 7 key compounds could be denoted as key contribution components for the bitterness index of 4 CTLs.

Table 4 VIP and P values of differential compounds in mainstream smoke of 4 CTLs.
Full size table

Correlation heat-map analysis

The correlation heatmap is known as a visualization method based on correlation coefficients between different variables, while the color gradients directly represent the correlation strength of the variables23. Pearson correlation analysis was conducted using the 33 preliminarily screened mainstream smoke compounds and the results were shown in Fig. 4. It is interesting to note that myosmine (2), 2,3’-bipyridine (4), cotinine (5), 3-ethyl-pyridine (10), phenol (19), phytol (25), and nicotinamide (27) exhibited highly significant positive correlations (P < 0.01) with bitterness taste.

Fig. 4
Fig. 4
Full size image

Correlation heatmap of mainstream smoke compounds of 4 CTLs. Red colors represent positive correlations and blue colors indicate negative correlations. Color depth and asterisk numbers of represent the strength of correlation, and deeper red colors and more asterisks indicate stronger relationships.

PLSR correlation analysis

To further investigate the relationship between volatile compounds in mainstream smoke and the bitterness tastes, PLSR analysis were performed using the Unscrambler software. The bitterness sensory evaluation scores were selected as the dependent variable (Y), and 33 preliminarily screened mainstream smoke compounds were used as explanatory variables (X). As shown in Fig. 5, the PLSR model could effectively explained correlations between sensory attributes (located between the two ellipses) and mainstream smoke compounds. Compounds that located closer to the bitterness taste vector in the model space can make greater positive contributions to this attribute24. It can be seen from Fig. 5 that myosmine (2), 2,3’-bipyridine (4), cotinine (5), 3-ethyl-pyridine (10), phytol (25), and nicotinamide (27) were close to the bitterness taste vector, which was also consistent with the above OPLS-DA screening and correlation heat map analysis results.

Fig. 5
Fig. 5
Full size image

PLSR analysis of mainstream smoke compounds of 4 CTLs with bitterness taste.

Thus, through comprehensive analysis of the results of OPLS-DA, Pearson correlation analysis, and PLSR model, myosmine (2), 2,3’-bipyridine (4), cotinine (5), 3-ethyl-pyridine (10), and nicotinamide (27) were revealed as the key and mutual mainstream smoke compounds associated with bitterness taste. Additionally, nicotine (1) was also considered as a candidate bitterness component due to its high concentration and low taste threshold. The chosen compounds were further verified by systematic sensory evaluation.

Smoke sensory verification of bitterness compounds

The above 6 chosen bitterness compounds were prepared into aqueous solutions to a concentration of 1000 µg/mL. Using test cigars rolled by Chuxue 14 CTLs, 20 µL of each solution was injected via central injection methodology, and control group cigars were injected with equal volumes of water. Then, the sensory evaluation was performed mainly focused on bitterness taste intensity. Compounds 2,3’-bipyridine, nicotinamide, nicotine, and myosmine brought significant bitterness taste enhancement (Fig. 6) with sensory evaluation scores from 6.5 to 8.8 points, while 3-ethyl-pyridine and cotinine exhibited weak effects (2.0 points respectively). These results suggested 2,3’-bipyridine, nicotinamide, nicotine, and myosmine as the key compounds associated with CTL smoke bitterness22.

Fig. 6
Fig. 6
Full size image

Bitterness sensory evaluation scores of test cigars injected with different compounds. CK, control group cigar made by Chuxue 14 CTLs, injected with water.

Taste verification of bitterness compound in aqueous solution

The selected four compounds with high bitterness intensity were prepared into aqueous solutions with serial concentration gradients. The taste sensory evaluation were carried out and the bitterness scores were calculated. It was found that the bitterness intensity of different compounds enhanced along with the increase of concentration, and final bitterness intensity achieved at higher concentrations (Fig. 7). The bitterness intensity of 2,3’-bipyridine and myosmine was strong, and relatively low concentrations (8 µg/mL for myosmine and 20 µg/mL for 2,3’-bipyridine) could bring the final high bitterness intensity. Nicotine possessed obvious spicy taste in addition to bitterness at a concentration of 150 µg/mL. And the final bitterness intensity of nicotinamide achieved 7.5 points at a concentration of 100 µg/mL.

Fig. 7
Fig. 7
Full size image

Bitterness taste sensory evaluation scores of 4 key compounds at different concentrations.

Molecular docking of key bitterness compounds with bitter taste receptors

A total of 25 hTAS2Rs have been identified and reported, which can bond with structurally diverse compounds related to bitterness taste. Among them, TAS2R1, TAS2R7, TAS2R8, TAS2R10, TAS2R14, and TAS2R46 are known as broad-spectrum bitterness taste receptor that can be activated by a variety of bitterness substances2. Moreover, the cryogenic electron microscopy structures of TAS2R14 has been identified, and its broad-spectrum ligand recognition mode was elucidated, which could discern over 150 distinct bitterness compounds ranging from food to medicines1,5. The sequences and structures of TAS2R9, TAS2R13, and TAS2R16 are predicted similar to that of TAS2R14. Therefore, the above 9 TAS2Rs (TAS2R1, TAS2R7, TAS2R8, TAS2R9, TAS2R10, TAS2R13, TAS2R14, TAS2R16, and TAS2R46) were selected for molecular docking, and the binding energy of four key bitterness compounds with these receptors was summarized and analyzed18. The intermolecular interactions and molecular dynamics simulations of four key bitterness compounds with TAS2R14 bitterness taste receptor were further investigated.

Fig. 8
Fig. 8
Full size image

Heatmap of binding energy (kcal/mol) distribution of 4 key bitterness compounds with 9 bitter taste receptors.

As shown in Fig. 8, 2,3’-bipyridine exhibited the overall strongest binding effects (average − 6.4 kcal/mol) with 9 receptors than the other 3 compounds, with a lowest binding energy of −7.5 kcal/mol with TAS2R14. Nicotine could be easily discerned by TAS2R16 and TAS2R14, while myosmine possessed strong binding energy with TAS2R10. It is generally known that a binding energy lower than − 5 kcal/mol represents strong docking activity20,25. Among the receptors, TAS2R14 possessed relatively sensitive ligand recognition ability with average binding energy of −6.3 kcal/mol with 4 bitterness compounds.

As shown in Fig. 9, the bitterness compounds primarily bound to the active sites of the bitter taste receptor TAS2R14 through hydrogen bond interaction, alkyl interactions, and π-alkyl interactions. Hydrogen bond interactions play a crucial role, since this kind of action possesses strong binding affinity26. Interestingly, both the formation of complexes of 2,3’-bipyridine and myosmine with TAS2R14 were mainly through conventional hydrogen bond site of SER-194 and π-alkyl interaction site of PHE-198 (Fig. 9A and D). And myosmine could also work with TAS2R14 through alkyl interaction site of VAL-233 (Fig. 9D). Moreover, the main docking sites for the complex of ligand nicotinamide with TAS1R14 were π-alkyl interaction sites of TYR-107 and PHE-198, while π-alkyl interaction site of TRP-89 and alkyl interaction sites of VAL-180 and LEU-85 were revealed for nicotine-TAS2R14 complex (Fig. 9B and C). Overall, the key binding sites for bitterness compounds with TAS2R14 were revealed as SER-194, PHE-198, TYR-107, and TRP-89.

Fig. 9
Fig. 9
Full size image

Molecular docking sites of key bitterness compounds with TAS2R14. (A) 2,3’-bipyridine; (B) Nicotinamide; (C) Nicotine; (D) Myosmine.

Molecular dynamics simulation results

The primary and sole criterion for using the TAS2R14 receptor to molecular dynamics (MD) simulations was its superior energetic profile. Specifically, TAS2R14 complexes with the four key bitterness compounds exhibited the strongest average binding energy of −6.3 kcal/mol.

To validate the reliability of the molecular docking results, molecular dynamics simulations were performed on the complexes with the root mean square deviation (RMSD) results as shown in Fig. 10. RMSD generally represents the positional deviation between the protein’s simulated conformations and its initial structure, which plays a crucial role for evaluating system stability. Lower and stabilized RMSD values typically suggest a stable protein-ligand complex conformation and reliable binding25,27. Throughout a 50-ns dynamics simulation, all four bitter compound-TAS2R14 systems maintained stable binding states. The simulation systems of 2,3’-bipyridine, nicotinamide, nicotine, and myosmine were gradually stable after around 10 ns, 40 ns, 40 ns, and 15 ns, respectively. The RMSD values ultimately converged to around 2.5 Å, which indicated system equilibrium.

Fig. 10
Fig. 10
Full size image

Root mean square deviation (RMSD) of the bitterness compound-TAS2R14 receptor complexes.

The root mean square fluctuation (RMSF) results of the simulated systems were presented in Fig. 11. RMSF is generally known as a key indicator to measure the fluctuation degree of protein residues during the simulation process, and lower value indicated the smaller conformational fluctuation in the region27. As shown in Fig. 11, all 4 simulated systems exhibited great fluctuations in 3 regions: VAL-70 to LYS-80, ARG-160 to PHE-175, and THR-215 to ALA-226. These simulated RMSF values suggest structural perturbations in these regions, but the key binding residues identified in the TAS2R14 binding pocket such as SER-194, PHE-198, TYR107, and TRP-89 were not included. The RMSF values of the above key binding residues were revealed as low throughout the simulations, which suggested the stable interactions of four bitterness compounds-TAS2R14 located in the binding pocket of the receptor.

Fig. 11
Fig. 11
Full size image

Root mean square fluctuation (RMSF) of bitterness compound-TAS2R14 receptor complexes.

Conclusion

This study systematically investigated the key bitterness compounds in CTLs and the possible interactions of key bitterness compounds with bitterness taste receptors. Four key compounds related with bitterness taste, including nicotine, 2,3’-bipyridine, myosmine, and nicotinamide, were comprehensively demonstrated through sensory evaluation, mainstream smoke compound analysis, OPLS-DA, heat-map analysis, and PLSR model. The results were further verified by sensory evaluation. Molecular docking revealed that the four compounds primarily interacted with bitterness taste receptor TAS2R14 through hydrogen bonding, π-alkyl, and alkyl interactions. Overall, the key binding sites for bitterness compounds with TAS2R14 were revealed as SER-194, PHE-198, TYR-107, and TRP-89. Molecular dynamics simulations further demonstrated the formation of stable complexes between these compounds with TAS2R14. This study provides a scientific insight into the key bitterness compounds of CTLs, which is of great importance for their quality control, taste regulation, and agricultural production. Further studies on exploring the formation mechanism of bitterness components and developing of special bitterness inhibitors for CTLs are needed.