Fusobacterium nucleatum ATCC 25586 and P. gingivalis ATCC 33277 were both obtained from the Industrial Microbial Culture Collection and Management Center of China, and routinely cultivated on brain heart infusion (BHI, Difco, MD, United States) medium containing 0.005% hemin chloride and 0.001% vitamin K. The origin of the Elsholtzia ciliata (Thunb.) plant is from Yichun (Jiangxi, China; geographic coordinates are 113°54′–116°27′ E, 27°33′–29°06′N). E. ciliate essential oil (abbreviated as ECO in the text) was prepared from the stems and leaves of E. ciliate using a steam distillation method as previously described (Aziz et al., 2018).

1 × 10⁸ cells/mL F. nucleatum and the 2 × 10⁸ cells/mL P. gingivalis diluted by BHI broth were, respectively, added to a 96-well tissue culture plate (Orange Scientific, BA, Belgium) with different concentrations of samples for testing. The bacterial cell concentrations were measured by reading OD_600 nm on a plate spectrophotometer (EZ Read 800 Plus, Biochrom, Cambridge, United Kingdom). The minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC) were determined as previously described (Chitsazha et al., 2022).

F. nucleatum (1 × 10⁸ cells/mL) or P. gingivalis (2 × 10⁸ cells/mL) was inoculated in BHI broth, and a 5 mL aliquot was cultivated in 10 mL sterile glass tubes using an anaerobic incubator (Nuoxuan, Shanghai, China) for 12 h, and the tubes were shaking vigorously during the incubation to prevent biofilms formation. CHX or ECO was added to the broth at the concentration of 1/2× MIC. After incubation, a Halimeter (Yuanxun, Guangdong, China) was employed to measure VSCs concentration and record the data according to a published protocol (Tanabe et al., 2012). The inhibition rate of VSCs production by CHX or ECO was calculated as the percentage reduction in VSCs production of the treated group compared to the untreated control group. Triplicate experiments were conducted.

Fusobacterium nucleatum (1 × 10⁸ cells/mL) or P. gingivalis (2 × 10⁸ cells/mL) was inoculated in BHI broth in a 24-well tissue culture plate (Orange Scientific) with different concentrations of CHX or ECO, incubated at 37°C for 12 h, and the plates were shaking vigorously during the incubation to prevent biofilms formation. The bacterial cell concentrations were measured by reading OD_600 nm at different time points to generate the growth curve.

Bacterial biofilms were cultivated and quantified as previously described (Aherne et al., 2022) with minor modifications. Fusobacterium nucleatum and P. gingivalis cells were mixed in BHI broth at a final concentration of 1 × 10⁸ cells/mL and 2 × 10⁸ cells/mL, respectively, and added to a 96-well plate for anaerobic incubation for 48 h. Remove the supernatant, refill the well with fresh BHI broth containing different amounts of ECO, and anaerobically incubated for 24 h. Remove the supernatant, stain the biofilms with 0.1% crystal violet, and wash with sterile PBS three times, add 200 μL of 33% acetic acid to the well. The biomass of biofilm was determined by measuring OD_590 nm on a plate spectrophotometer. The biofilm elimination rate of ECO was calculated as the percentage reduction in biomass between the treated group and the untreated group. Triplicate experiments were conducted.

The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT, Solarbio, United States) colorimetric assay (He et al., 2022) was employed to determine the metabolic activity of biofilms. Briefly, the 48 h-biofilms were prepared as described above, after removing the supernatant, the biofilms were washed three times with sterile PBS, and 200 μL PBS containing different amounts of ECO was refilled and incubated anaerobically for 4 h. 90 μL BHI and 10 μL of 5 mg/mL MTT were added to each well, and incubated at 37°C in the dark for 2 h. After incubation, discard the supernatant and wash the well with PBS, and add 200 μL DMSO to each well. The OD_490 nm was read on a plate spectrophotometer. The biofilm metabolic inhibition rate was calculated as the percentage reduction in OD_490 nm value between the treated group and the untreated group. Triplicate experiments were conducted.

The biofilm formed by Fusobacterium nucleatum and P. gingivalis was cultured in a confocal laser petri dish (Solarbio, Beijing, China) as described above. After 48 h incubation, discard the supernatant and wash the well with PBS, and add PBS containing ECO or CHX at a concentration of 4× MIC, respectively. After incubation at 37°C for 4 h, the biofilms were stained SYTO 9 and propidium iodide (PI) within the LIVE/DEAD viability kit (Thermo Fisher, MA, United States) following the producer’s instruction. An LSM880 confocal laser scanning microscopy (CLSM) (Zeiss, Oberkochen, Germany) was used to observe and photograph the data through a 40× objective lens. The CLSM images were exported in an 8-bit TIFF format and statistically analyzed using the ImageJ software (Memmel et al., 2020). At least three random fields of each biofilm sample were examined.

The Q Exactive GC Orbitrap GC-MS/MS system equipped with a DB-5 MS U1tra Inert weakly polar column (30 m, 0.25 mm, 0.25 μm; Agilent, Santa Clara, CA, United States) was utilized for gas chromatography-mass spectrometry (GC-MS) analysis. The electron ionization (EI) source was employed and the interface temperature was set at 230°C. The full scan mode was utilized for data acquisition with the m/z range of 40–500 amu. The incubation temperature was 37°C with a 30 min incubation time, carrier gas flow rate was 1 mL/min, and injection temperature was 200°C. The temperature program for the GC analysis was as follows: started at 35°C for 6 min, ramped to 150°C at 4°C/min and held for 3 min, further ramped to 250°C at 40°C/min and held for 5 min, resulting in a total runtime of 45 min. The relative percent content of each component was determined by normalizing the peak areas of the TIC chromatogram. The mass spectrum was utilized to identify the structures of the components by searching NIST20 and WILEY275 databases and comparing with authentic standard spectra. The main chemical components of BCO were determined and quantified using direct liquid injection mode.

For biofilm VSCs analysis, 5 mL of BHI broth containing a concentration of 1 × 10⁸ cells/mL F. nucleatum or 2 × 10⁸ cells/mL P. gingivalis was added to sterilized headspace injection vials, respectively. The vials were then anaerobically incubated at 37°C for 12 h. In the drug treatment group, BCO was added at a final concentration of 2× MIC, whereas no drugs were added in the control group, and BHI medium without bacterial cells was used as a negative control. Gas composition in the headspace of the vials was analyzed using GC-MS with headspace injection mode (Milanowski et al., 2019), and the parameters are the same as described above.

According to the GC-MS results, carvacrol, p-cymene, and phellandrene were identified as the major chemical components in ECO, therefore, the standard compounds of carvacrol, p-cymene and phellandrene were obtained from Macklin Biochemical Co. (Shanghai, China) for further experiments. MIC and MBC of the compounds against F. nucleatum and P. gingivalis planktonic cells were determined as mentioned above. Furthermore, using a checkerboard assay, the antibacterial effect of compound combinations was evaluated by performing concentration gradient joint sensitivity experiments as previously described (Priya et al., 2021). The judgment criterion for the sensitivity experiment of combination drug therapy is based on the fractional inhibitory concentration index (FICI). FICI ≤0.5 indicates synergistic effect, 0.5 < FICI ≤ 1.0 indicates additive effect, 1 < FICI ≤ 2 indicates irrelevant effect, and FICI>2 indicates antagonistic effect.

The cysteine stimulation test and methanethiol stimulation test were used to detect hydrogen sulfide and methanethiol in VSCs according to a published protocol (Liu P. F. et al., 2013) with minor modifications. To determine the concentration of hydrogen sulfide, the experimental setup was the same as that of MIC determination, and 10 μL of 0.6% cysteine solution and 10 μL of 2.4% lead acetate solution were added to each well. The plate was incubated anaerobically at 37°C for 12 h, the tubes were shaking vigorously during the incubation to prevent biofilm formation, and OD_550 nm was measured on a plate spectrophotometer. To determine the concentration of hydrogen sulfide, 10 μL of 1.8% methionine solution and 10 μL of 0.18% DTNB solution were added to each well, and OD_430 nm was measured. Triplicate experiments were conducted. The VSCs production rate was calculated as the percentage of hydrogen sulfide and methanethiol production in the treated group versus that of the untreated group. Triplicate experiments were conducted.

F. nucleatum and P. gingivalis cells were harvested, washed with PBS buffer for three times, resuspended in PBS with or without 2× MIC ECO, and incubated at room temperature for 4 h. The cells were collected by centrifugation and fixed with 2.5% glutaraldehyde (Thermo Fisher) for 3 h, and subsequently dehydrated using different concentrations of ethanol. The specimen was observed through a scanning electron microscope (Inspect F50, FEI, United States).

F. nucleatum and P. gingivalis cells were collected by centrifugation, resuspended in PBS with or without 2× MIC ECO or CHX to a final concentration of 1 × 10⁸ cells/mL and 2 × 10⁸ cells/mL, and incubated in anaerobic incubator for 10 h. For every 2 h, a small aliquot was taken and stained with PI for 15 min under dark conditions. The fluorescence intensity was measured using a fluorescence microplate reader (TECAN, Shanghai, China), with excitation at 490 nm and emission at 635 nm. Triplicate experiments were conducted.

The saliva-derived bacterial biofilms containing halitosis-causing bacteria were cultivated according to a published method (Xu et al., 2022) with modifications. Ten healthy volunteers from Shandong University were selected, who had not taken any proton pump inhibitors, antibiotics, or corticosteroid medications, and had no other oral diseases, such as dental caries, periodontitis, or oral ulcers. Pooled saliva samples were collected immediately after waking up in the morning, added to a BHI medium, and incubated under microaerobic conditions at 37°C for 12 h. F. nucleatum and P. gingivalis were added in the broth at a final concentration of 5 × 10² cells/mL and 1 × 10³ cells/mL, respectively. The broth was further incubated anaerobically at 37°C for 36 h to obtain the saliva-derived bacterial biofilm.

For the treatment experiment, CHX or ECO was added to the saliva-derived bacterial biofilm after 36 h incubation, e.g., 24 h after the addition of F. nucleatum and P. gingivalis, to a final concentration of 1× MIC, and BHI liquid was added to the control group. After cultivation for additional 12 h, the bacterial cells were collected and stored at −80°C before further testing.

Total genomic DNA was extracted from nine biofilm samples using the TGuide S96 Kit (Tiangen, Beijing, China) according to the manufacturer’s instructions. The hypervariable region V3-V4 of the bacterial 16S rRNA gene was amplified with primer pairs 338F: 5′-ACTCCTACGGGAGGCAGCA-3′ and 806R: 5′-GGACTACHVGGGTWTCTAAT-3′. PCR products were checked on agarose gel and purified through the Omega DNA purification kit (Omega, GA, United States). The purified PCR products were collected and the paired ends (2 × 250 bp) were performed on the Illumina Novaseq 6000 platform in Biomarker Technologies Co. (Beijing, China).

The qualified sequences with more than 97% similarity thresholds were allocated to one operational taxonomic unit (OTU) using USEARCH (version 10.0) (Edgar, 2013). Taxonomy annotation of the OTUs was performed based on the Naive Bayes classifier in QIIME2 (Davis et al., 2019) using the SILVA database (Quast et al., 2013) with a confidence threshold of 70%. Alpha and beta diversity analysis was performed to identify the complexity of species diversity of each sample utilizing QIIME2 software. Distance matrices were obtained by the Bray–Curtis algorithm. Principal coordinates analysis (PCoA) was generated by the R package. Hierarchical clustering of samples was performed by an unweighted pair-group method with arithmetic mean (UPGMA) to determine the similarity in species composition among samples. One-way analysis of variance was used to compare bacterial abundance and diversity. The online platform BMK Cloud was used to analyze the sequencing data.¹

GraphPad Prism version 7 (La Jolla, CA, United States) was used for statistical analysis of the data. The quantitative data met the normal distribution and the homogeneity of variance, expressed as mean ± SD. One-way ANOVA was used for comparison between groups unless noted otherwise.

The in vitro inhibitory and killing activity of ECO against planktonic cells of F. nucleatum and P. gingivalis was determined using the serial dilution method. As shown in Figures 1A,C, the MIC and MBC values of ECO for both bacteria were similar, indicating that its antibacterial activity is mainly bactericidal rather than bacteriostatic. Consistent with this observation, ECO at concentrations higher than the MIC significantly inhibited bacterial proliferation in the liquid medium, resulting in growth curves similar to those of the CHX-treated group (Figures 1B,D). When the ECO concentration was lower than the MIC, the growth of both bacteria was similar to that of the untreated group. Meanwhile, ECO showed slightly better bactericidal and growth-inhibiting activity against planktonic cells of both bacteria than CHX, demonstrating its good potential in applications. It is worth noting that since ECO is a liquid sample, the concentration units of its MIC and MBC are different from CHX. However, after performing a density conversion, the same conclusion can still be obtained. For example, the MIC of ECO against F. nucleatum is 0.1 μL/mL, which is approximately 2.0 μg/mL after conversion, similar to the MIC value of CHX. At sub-lethal concentrations, i.e.,1/2× MIC, ECO and CHX did not affect the growth of both bacteria (Figures 1B,D), while significantly reducing their VSCs production by approximately 60% during the anaerobic liquid culture process (Figures 1A,C). This result indicates that even if ECO does not have a killing effect on the cells of both bacteria, it can significantly inhibit their gas-producing metabolic processes.

Various bacteria can form mixed biofilms and colonize in the oral cavity, and bacteria in biofilms normally exhibit greater tolerance to the drug treatments versus their planktonic growth state (Wang et al., 2017). Therefore, we cultured mixed biofilm of F. nucleatum and P. gingivalis, and determined the effect of ECO on its elimination, metabolic inhibition, and cell-killing ability. As shown in Figure 2A, ECO exhibited a significant dose-dependent removal effect on the mature mixed biofilms, which is slightly better than that of CHX. At 1× MIC, ECO eliminated approximately 29.10% of the biofilm, and the clearance rate increased to approximately 50% at concentrations from 4× to 16× MIC. Similar to CHX, ECO showed a significant metabolic inhibitory effect on bacterial cells in the mixed biofilm, achieving a close to 35% inhibition rate even at a lower concentration (1× MIC), which was consistent with the results obtained in planktonic cells as mentioned above.

To investigate the killing effect of ECO on bacterial cells in the biofilm, CLSM was employed to observe the biofilms visualized by live-dead staining assay after drug treatment. According to the biofilm elimination results (Figure 2A), a higher concentration (4× MIC) of CHX or ECO was used in the experiment. As shown in Figure 2B, compared with the untreated group, both CHX and ECO treatments remarkably increased the red fluorescence signal and decreased the green signal in the mixed biofilm, indicating a relatively significant increase in the number of dead cells in the biofilm due to a definitive bactericidal activity. By calculating the ratio of dead cells, at a concentration of 4× MIC, ECO had a relatively superior killing effect on bacterial cells within biofilm compared to CHX. Moreover, this result also confirmed that F. nucleatum and P. gingivalis cells in the mixed biofilm exhibit significantly increased resistance to various drugs compared to their planktonic growth state.

The chemical composition of the prepared ECO was analyzed using GC-MS. A total of 137 distinct peaks were identified in the obtained total ion chromatogram (TIC, Figure 3A and Supplementary Table S1), with the highest peak intensities of the main components concentrated between 9–21 min of retention time. The relative percentage content of each peak was calculated using the area normalization method (Supplementary Table S1). By searching the MS database and matching with standard spectra, a total of 28 compounds were identified, accounting for 97.10% of the total peak area. Among these compounds, carvacrol (Figure 3B), p-cymene (Figure 3C), and phellandrene (Figure 3D) were the major chemical components of ECO, accounting for 67.83% of the total peak area. These results are consistent with previous reports (Phetsang et al., 2019; Liang et al., 2021).

MIC and MBC results for carvacrol, p-cymene, and phellandrene showed good in vitro inhibitory and bactericidal activities against both F. nucleatum and P. gingivalis planktonic cells (Figure 4A). Among them, carvacrol had the best activity, with slightly bigger MIC than ECO, while the other two compounds showed significantly lower activity than ECO. Considering that carvacrol accounts for only approximately 26.25% of ECO (Supplementary Table S1), the overall activity of ECO may be derived from the cumulative effect or synergistic action of its multiple active components. Therefore, the planktonic cells of F. nucleatum and P. gingivalis were treated with the combination of the three compounds (Figure 4B), and FICI values (Table 1) were used to assess the inhibitory effects of the compound combination. The results showed that when the three compounds were combined in pairs, they exhibited certain synergistic inhibitory effects on F. nucleatum, and good additive inhibitory effects on P. gingivalis, thereby further verifying our hypothesis.

Cysteine and methionine can be rapidly metabolized by oral halitosis-causing bacteria into malodorous gases such as hydrogen sulfide and methanethiol, which can trigger bad breath (Salako and Philip, 2011). Therefore, we conducted gas production stimulation tests on planktonic cells of F. nucleatum and P. gingivalis by adding L-cysteine and L-methionine in the medium. As shown in Figure 5A, the production of hydrogen sulfide and methanethiol by both bacteria was almost completely suppressed when the concentration of ECO was at the MIC level. At sub-lethal concentrations, ECO also partially inhibited the gas production process, which is consistent with our previous results (Figures 1A,C). Compared with the untreated group, ECO was more effective at inhibiting gas production in F. nucleatum than P. gingivalis, and its inhibitory effect on hydrogen sulfide production in F. nucleatum was the most significant.

Furthermore, the chemical composition of the gas produced during the biofilm formation of F. nucleatum and P. gingivalis was determined using GC-MS (Figures 5B,C). The major components in the gas produced by F. nucleatum biofilm were hydrogen sulfide, dimethyl disulfide, methanethiol, alanine ethyl ester, and methyl thioacetate, which accounted for 97.61% of the total (Supplementary Table S2). Treatment with 2× MIC ECO significantly diminished the gas production of F. nucleatum (Figure 5B). Except for the volatile components from ECO, only a small amount of hydrogen sulfide was detected as the main component of VSCs after the treatment, which decreased by about 92% compared to the untreated group (Supplementary Tables S2, S4). The gas produced by P. gingivalis biofilm was mainly composed of hydrogen sulfide, dimethyl disulfide, and methanethiol, which accounted for 96.99% of the total (Supplementary Table S3). Treatment with 2× MIC ECO significantly inhibited the gas production of P. gingivalis (Figure 5C). Hydrogen sulfide was also detected as the main component of VSCs, which decreased by about 76% compared to the untreated group (Supplementary Tables S3, S5).

After ECO treatment for 4 h, changes in cell morphology of F. nucleatum and P. gingivalis were examined using scanning electron microscopy (SEM). As shown in Figure 6A, untreated F. nucleatum cells were plump and elongated with a regular spindle shape, tapered at both ends, and slightly swollen in the middle with smooth edges. After ECO treatment, the cells exhibited fragmentation and were accompanied by ruptures and damage to the edges. Untreated P. gingivalis cells were short rods with smooth and plump surfaces, while after ECO treatment, the cells lost their original bacterial morphology and underwent significant fragmentation. This phenomenon is similar to the CHX treatment results reported previously (Shen et al., 2016), suggesting that ECO is likely to exert a detrimental impact on the bacterial cell membrane. This can occur either through direct interaction with the membrane or by targeting other intracellular components, ultimately resulting in cell lysis.

To further verify this hypothesis, F. nucleatum and P. gingivalis cells were stained with PI to quantitatively measure the degree of their membrane damage caused by ECO treatment, and CHX was used as a positive control. As shown in Figure 6B, compared with the untreated group, both ECO and CHX increased the cell membrane permeability of both two bacteria, and this effect became more pronounced with prolonged exposure time. Furthermore, immediate membrane damage was observed upon the addition of both CHX and ECO, suggesting that it could be attributed to the direct interaction between these compounds and the cell membrane. Overall, ECO and CHX had a stronger ability to damage the cell membrane of F. nucleatum than P. gingivalis, which is consistent with our SEM results (Figure 6A). In the shorter treatment time, the effect of ECO on F. nucleatum was similar to that of CHX, while CHX was more effective than ECO on P. gingivalis. After 10 h of treatment, the destructive effect of ECO on the cell membrane of both bacteria was significantly higher than that of CHX (p < 0.01).

The raw sequence data was processed as described above, after quality control and assembly, the clean reads from the control group were 77,049 ± 4,553, 79,694 ± 116 for the CHX group, and 79,519 ± 160 for the ECO group. The Shannon’s index was employed to access the Alpha diversity of the samples, which reflects species richness and community evenness. The index values did not have significant differences among the three groups, while a decrease was found both in CHX and ECO-treated groups compared to the control group (Figure 7A), and the index of ECO group was slightly bigger than that of CHX group. The result showed that the bacterial diversity in the untreated samples was higher than that in the treated samples. Regarding Beta diversity analysis, PCoA analysis was performed to explain the relationships between the bacterial composition of different biofilm samples, and the plot showed that the samples of three different groups were clustered separately (Figure 7B). This result revealed that microbiota in the saliva-derived bacterial biofilm with and without treatment was significantly different, and the difference in microbiota after treatment was also significant for groups CHX and ECO. It was suggested that CHX and ECO treatments altered biofilm microbiota, and the alternation was strong enough to separate the microbiota among different groups, which was confirmed by the constructed UPGMA clustering tree (Figure 7C, left panel).

At the genus level, 96.90% of the reads in all samples were classified into 9 known genera (Figure 7C, right panel). The ten most abundant genera (except the unclassified) in the control, CHX, and ECO group accounted for 94.89%, 97.35%, and 98.45%, respectively. Next, the abundance of bacteria related to oral diseases in biofilms was analyzed. As shown in Figure 7D, the average relative abundances of F. nucleatum than P. gingivalis in the untreated samples were 0.0628 ± 0.0398%, and 5.7402 ± 2.8532%, respectively; in CHX treated samples were 0.0271 ± 0.0030%, and 0.0325 ± 0.0325%, respectively; and in ECO treated samples were 0%, and 0.0103 ± 0.0010%, respectively. Similarly, bad breath-related Prevotella, Porphyromonas, Cryptobacterium and Stomatobaculum also showed a significant decrease in relative abundance in the treated samples, and the effect of ECO was slightly better than that of CHX.