L. angustifolia linalool synthase (LaLINS), limonene synthase (LaLIMS), and bergamotene synthase (LaBERS) are terpene synthases characterized by the presence of the following two conserved domains: the terpene synthase domain and the terpene cyclase domain ( Figure 1 ; Supplementary Figure S1 ). The molecular weights of LaLINS, LaLIMS, and LaBERS were approximately 65.65, 70.35, and 62.41 kDa, respectively. Their molecular formulas were determined to be C₂₉₇₉H₄₅₂₈N₇₇₂O₈₇₄S₁₆ for LaLINS, C₃₁₆₇H₄₉₀₃N₈₄₇O₉₂₇S₂₀ for LaLIMS, and C₂₇₉₈H₄₃₁₈N₇₃₂O₈₃₁S₂₈ for LaBERS. The isoelectric points (pI) were calculated as 5.08 for LaLINS, 5.49 for LaLIMS, and 5.16 for LaBERS, with corresponding protein instability indices of 48.94, 43.56, and 45.59, respectively.

The three-dimensional (3D) structures of LaLINS, LaLIMS, and LaBERS were predicted using Alphafold2 (Jumper et al., 2021; Tunyasuvunakool et al., 2021; Wayment-Steele et al., 2023) ( Figure 2 ). Unlike conventional homology modeling techniques, Alphafold2 utilizes advanced deep learning algorithms, which significantly improve the accuracy and reliability of protein structure predictions ( Supplementary Figures S2–S5 ).

To assess the predicted structural models, the Ramachandran Plot was used to assess the dihedral angles of the protein backbone ensuring the conformational integrity of the proteins. The structural models of LaLINS, LaLIMS, and LaBERS demonstrated that 93.6%, 88.2%, and 93.9% of their residues, respectively, were located within the most favored regions ( Figures 2B, D, F ; Table 1 ). Additionally, 6.4%, 9.7%, and 6.1% of residues were found in the allowed regions, while 0%, 1.4%, and 0% were situated in the generously allowed regions ( Figures 2B, D, F ; Table 1 ). These findings indicate that the structural models of the three enzymes are of high quality.

To investigate the oligomeric state of the three terpene synthases (LaLINS, LaLIMS, and LaBERS), dynamic light scattering (DLS) experiments were performed following centrifugation to measure their hydrodynamic radii. The results revealed hydrodynamic radii of 5.7 ± 0.2 nm for LaLINS, 6.2 ± 0.3 nm for LaLIMS, and 5.4 ± 0.2 nm for LaBERS ( Figure 3 ) showing that all three terpene synthases exist in monomeric forms.

Based on the sequence alignments of terpene synthases ( Figure 4 ; Supplementary Figure S1 ), targeted mutation experiments were conducted. The results showed that the mutations R283A, R461A, and T468A in LaLINS resulted in a two- to fivefold decrease in activity compared to the wild-type (WT) protein, whereas the mutations D320A and D324A completely abolished the activity ( Figures 4 , 5A ). Similarly, the mutations R319A, R497A, and T504A in LaLIMS led to a 5- to 20-fold reduction in the activity, while D356A and D360A entirely eliminated its activity ( Figures 4 , 5B ). For LaBERS, the mutations R254A, R432A, and T439A caused a two- to sevenfold decrease in the activity, with D291A and D295A resulting in a complete loss of function ( Figures 4 , 5C ). These findings highlight the critical role of negatively charged aspartic acid residues in protein function, particularly their proximity to the substrate, which underscores the importance of these mutations. The substrate likely fits within the binding pocket of the terpene cyclase domain through electrostatic interactions involving magnesium ions (Mg²⁺) and the side chains of aspartic acids. These results demonstrated that conserved negatively charged sites are essential for the terpene synthase activity of LaLINS, LaLIMS, and LaBERS, consistent with previous studies (Starks et al., 1997; Noel et al., 2010).

To evaluate the expression levels of the three terpene synthase genes (LaLINS, LaLIMS, and LaBERS) and the corresponding accumulation of metabolites at various times and in different tissues, real-time quantitative polymerase chain reaction (RT-qPCR) analysis was conducted using gene-specific primers ( Supplementary Table S1 ). We found that the expression of LaLINS, LaLIMS, and LaBERS was significantly higher in leaves compared to that in stems and flowers, with peak expression occurring at 8:00 a.m. ( Figure 6 ). These findings suggested that certain terpenes associated with LaLINS, LaLIMS, and LaBERS are synthesized and emitted from specific plant tissues at particular times, corresponding to the spatiotemporal expression patterns of their respective terpene synthases ( Figure 6 ). This implied that terpenoid biosynthesis is regulated at the transcriptional level. Leaf and stem tissues were primarily associated with secretory structures that produced large quantities of terpenes, predominantly monoterpenes and sesquiterpenes.

To further explore the spatiotemporal patterns of gene expression, gas chromatography-mass spectrometry (GC-MS) was employed to analyze the metabolites produced by the three terpene synthases. The results indicated that the highest metabolite yield occurred in the leaves compared to the corresponding yields from the other tissues (stem and flower) ( Table 2 ). Furthermore, the metabolite yield from the three terpene synthases at 8:00 a.m. was significantly greater than that observed at other time points (2:00 a.m., 2:00 p.m., and 8:00 p.m.) ( Table 2 ). These findings are consistent with the results obtained from RT-qPCR analysis.

These results are consistent with previous studies indicating that some terpene synthases show preferential expression in leaves (Dornelas and Mazzafera, 2007; Hamberger et al., 2020).

The procedures for constructing, expressing, and purifying various protein constructs of the three terpene synthases were conducted in accordance with established methods reported previously (Landmann et al., 2007; Liu et al., 2025).

The two-dimensional (2D) structures of the three terpene synthases were analyzed using the Phyre2 tool (Kelley et al., 2015). The three-dimensional (3D) structure predictions of these enzymes were carried out using the AlphaFold2 program (Jumper et al., 2021; Tunyasuvunakool et al., 2021; Wayment-Steele et al., 2023). Sequences for the terpene synthases were retrieved from the UniProt database (LaLINS, entry ID Q2XSC5; LaLIMS, entry ID Q2XSC6; LaBERS, entry ID Q2XSC4). Structural visualizations were generated using PyMOL 2.3.4 (https://www.pymol.org/2/). The quality of the structural model was evaluated through protein tertiary structure visualization and conformation charts produced by Ramachandran Plot analysis using PDBsum.

To investigate the oligomeric state of the three terpene synthases, their diameters were determined using dynamic light scattering (DLS) with a Dynapro DLS instrument (Malvern Zetasizer, Malvern, UK) following slight modifications (Liu et al., 2024). Each enzyme was concentrated to approximately 1.8 mg/ml and then subjected to centrifugation at 12,000 rpm at 4°C for 5 min. The enzymes were subsequently introduced into 1-cm path length cuvettes. Data acquisition involved performing 30 runs per sample, with an equilibration period of 120 s. The DLS data were analyzed using Zetasizer software (Version 6.20), and regularized DLS histograms were generated. The diameter of the particles was continuously monitored throughout the analysis.

The activities of the three terpene synthases were measured using a previously described method with some modification (Landmann et al., 2007). Standard assays were conducted in a total reaction volume of 500 μl, which included a buffer solution (25 mM Tris-Cl, pH 7.5, 5% glycerol, 1 mM DTT), 50 μM substrate (geranyl, farnesyl, or geranylgeranyl diphosphate), and 2–20 μg of purified recombinant enzyme. The reaction mixture was layered with 500 μl of diethyl ether and incubated at 23°C for a duration of 10 to 15 min under optimal conditions (Landmann et al., 2007). The reaction was terminated by vigorous mixing followed by centrifugation to facilitate phase separation. An internal standard was subsequently introduced (1 μg camphor for LaLIMS and LaBERS, and 0.164 μg [1,2-²H₂]-linalool for LaLINS), and the upper solvent phase was collected. A second extraction with an additional 500 μl of diethyl ether was performed. The combined extracts were concentrated to approximately 300 μl under a nitrogen stream, dried with sodium sulfate (Na₂SO₄), and subsequently analyzed using GC-MS (gas chromatography-mass spectrometry).

GC-MS (gas chromatography-mass spectrometry) analysis was conducted using an Agilent GC 6850 gas chromatograph coupled with an Agilent 5973 ion trap mass detector following a previously established method with some modifications (Landmann et al., 2007). The system was equipped with a 30 m × 0.25 mm apolar capillary column (DB5). The injector and detector temperatures were maintained at 250°C. Helium served as the carrier gas, delivered at a flow rate of 1.0 ml/min. The oven temperature program involved an initial hold at 60°C for 4 min, followed by a temperature ramp of 4°C/min until reaching 240°C, where it was held for an additional 5 min. A 2-μl injection was performed in splitless mode. Molecular identification was achieved using mass spectra databases, including Wiley, NIST 05, and Adams. The identification of linalool, limonene, and alpha-bergamotene was based on comparisons of the GC-MS data with those of authentic reference compounds.

Primers for site-directed mutagenesis of the target proteins were designed by our lab and synthesized by Shanghai Sangon Biotechnology (China) ( Supplementary Tables S5–S7 and Supplementary Figures S6–S8 ). Q5 polymerase from NEB (New England Biolabs) was used for PCR. Sequencing of the DNA was conducted by Shanghai Sangon Biotechnology (Shanghai, China) to confirm the accuracy of cloning sites. The procedures for expressing and purifying the mutated terpene synthases were identical to those used for the wild-type (WT) protein (Landmann et al., 2007). Additionally, the enzymatic activities of the mutated terpene synthases were evaluated under conditions equivalent to those used for the WT protein.

To quantify the expression levels of the three terpene synthases, real-time quantitative polymerase chain reaction (RT-qPCR) was performed using PowerUp SYBR Green Master Mix (Applied Biosystems). Total RNA was extracted with the Universal Plant Total RNA Extraction Kit (Bioteke, Beijing, China), following the manufacturer’s instructions. cDNA was subsequently synthesized from the RNA samples using the PrimeScript 1st Strand cDNA Synthesis Kit (Takara, Kyoto, Japan). Primers used for the analysis are detailed in Supplementary Table S1 . RT-qPCR was conducted with an Applied Biosystems QuantStudio 5 instrument. Data analysis was performed using the 2^−ΔΔCT method (Rocha et al., 2015; Biassoni and Raso, 2022; Hajeri et al., 2022; Liu et al., 2024), with relative expression ratios presented as log₂ values in histograms. Gene beta-actin (entry ID A0A2I8B2D2) served as the housekeeping gene for data normalization ( Supplementary Figure S9 ), and a positive control using the beta-actin gene was included in the analysis.

All experiments were conducted at least in triplicate. The data were expressed as mean ± SD. Statistical analysis was conducted using Origin 8.5, Microsoft Excel 2013, and SPSS 19.0. In the all statistical evaluations, p < 0.05 was considered statistically significant, and p < 0.01 was considered high statistically significant.