S. mutans (ATCC 251175) was obtained from the Guangdong Microbiology Culture Center and was cultured in brain heart infusion (BHI) broth (Hopebio, Qingdao, China) supplemented with 1% sucrose at 37°C for 24h under aerobic conditions. After incubation, the bacterial concentration was 10⁷cfu/ml, as determined by spectrophotometry (OD₆₃₀=0.2). Ursolic acid was purchased from the National Institutes for Food and Drug Control with purity >98% and was dissolved in dimethyl sulfoxide.

The potential inhibitory activity of ursolic acid against S. mutans was determined by the microdilution method as described previously (Liu et al., 2017), with minor modifications. Briefly, ursolic acid was serially diluted twofold in BHI broth containing 1% sucrose, with the final concentration of ursolic acid ranging from 0.25 to 0.031mg/ml. Twenty microliters of sterile solution of resazurin sodium per well was added to the bacterial culture and incubated at 37°C for 24h. BHI broth alone was used as blank control, bacterial suspension alone was used as a noninhibition negative control, and chlorhexidine treatment at a final concentration of 0.6mg/ml was used as a positive control. As bacteria grow, resazurin is reduced to resorufin, resulting in the medium color changing from blue to pink (Palomino et al., 2002; Sarker et al., 2007). The lowest concentration of ursolic acid that could inhibit the medium color change from blue to pink was defined as the minimal inhibitory concentration (Palomino et al., 2002; Süntar et al., 2016). The bacterial cultures treated with ursolic acid at concentrations equal to or higher than the minimum inhibitory concentration (MIC) were transferred to BHI agar plates and incubated at 37°C for 24h. The lowest concentration that resulted in no visible bacterial colonies on the agar plates after incubation was defined as the minimal bactericidal concentration (MBC; Liu et al., 2017).

The effect of ursolic acid on the viability of biofilms was evaluated using a 2,3-bis (2-methyloxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT) reduction assay as described previously (Duque et al., 2017), with some modification. Briefly, after 24h of incubation for cultured bacteria to form biofilms in 96-well microplates, supernatants were removed, and the wells were gently washed three times with PBS (pH 7.0) to remove nonadhered bacteria. Then, the cultures were incubated at 37°C for another 22h in 200μl of BHI broth including 100μl of ursolic acid at various concentrations. After removing the supernatants, 50μl of XTT (Sigma, St. Louis, MO, United States) reagent was added, and the microplate was kept in the dark for 2h at 37°C. Then, the absorbance of the colored product was detected at 490nm using an ELISA reader. Alternatively, after ursolic acid treatment, a crystal violet assay was used to evaluate the attachment of biofilm biomass as described previously (Cardoso Sá et al., 2012).

Fluorescence staining was used to determine biofilm integrity as described previously (Liu et al., 2017). After biofilm formation, ursolic acid was added at a final concentration of 1/2 MIC followed by incubation at 37°C for another 18h. The culture was then removed from the supernatant and gently washed twice with sterile water. Staining was carried out by means of the Live/Dead BacLight Bacterial Viability Kit (L13152, Invitrogen, Carlsbad, CA, United States) for 30min at room temperature in the dark. Stained cells were observed under a fluorescence microscope (Nikon Eclipse 80i; Nikon Co., Japan). The kit utilizes a mixture of SYTO 9, a green-fluorescent nucleic acid stains and propidium iodide (PI), a red-fluorescent nucleic acid stains. SYTO 9 is a membrane-permeable fluorescent marker that stains all cells, while PI penetrates only bacteria with damaged membranes and stains dead cells red. Twenty fields per sample were randomly selected to analyze the intensity of red and green fluorescence.

The method for quantification of EPS production was conducted as described previously (Packiavathy et al., 2012). Briefly, the bacterial culture was incubated with or without ursolic acid at concentrations ranging from 0.008 to 0.125mg/ml at 37°C for 16h, the culture was centrifuged (4°C, 12000×g, 30min), and the supernatants and cells were collected. The water-soluble and water-insoluble glucans were prepared using the method described previously (Yano et al., 2012). Briefly, water-soluble glucans were obtained by ethanol precipitation of the supernatants. The cells were resuspended in 1M NaOH and centrifuged to collect supernatants, which were used to prepare water-insoluble glucans by ethanol precipitation. The phenol/H₂SO₄ method was used to quantify two types of EPS in the supernatant as described previously (Liu et al., 2017). The absorbance of the color was detected at 490nm using an ELISA reader.

The crystal structure of glucansucrase was downloaded from the Protein Data Bank (ID: 3AIC) as the receptor, and the preparation work was done using the software Gold 5.2 and AutoDockTool 1.5.6. The MD simulation of the system was performed for 20ns under the npt ensemble, and the data were saved every other 5ps. CPPTRAJ was used for data analysis. The MMPBSA.py module was used to compute the binding free energy between the protein and ligand.

To further confirm the binding site between GTF-SI and ursolic acid, the amino acids in the predicted binding site were replaced by other amino acids, as shown in Supplementary Table S1 Afterward, the mutant GTF-SI proteins were subjected to molecular docking under the same conditions as above.

The method for analyzing the enzymatic activity of the crude extract of GTFs was used as described previously by Koo et al (Koo et al., 2000). A 20-mL bacterial suspension of S. mutans was incubated in 200ml of BHI broth containing 1% sucrose at 37°C. After incubation, supernatants were collected, and ammonium sulfate was added at 60% saturation for 24h to prepare the protein. The crude enzymes were dissolved in PBS (pH 6.0). The reaction system mixture was the same as that described previously (Liu et al., 2017). The final concentration of ursolic acid ranging from 0.04 to 0.12mg/ml was used to measure the inhibition of the synthesis of EPSs. After incubation, the method to determine the amount of the two types of glucans was applied as described above.

cDNA fragments of wild-type GTF-SI (GenBank: M22054.1) and two mutant GTF-SI variants (Supplementary Table S1) were synthesized and inserted into the NdeI and XhoI sites of pCold I to generate pCold-GTF-SI recombinant plasmids with a 6× histidine tag at the N-terminus of the proteins (Ito et al., 2011). The recombinant proteins were purified by using GE metal affinity resin and dialyzed. SDS-PAGE was used to analyze recombinant GTF-SI proteins (shown in Supplementary Figure S1).

Capillary electrophoresis (CE) was used to quantify fructose production catalyzed by GTFs as described previously (Rizelio et al., 2012). Recombinant GTF-SI was crudely extracted from recombinant transformants by ultrasonic disruption. Then, 50μl of the crude enzyme solution together with or without ursolic acid at a final concentration of 0.03mg/ml or 0.02mg/ml was added to 1ml of sucrose (0.1M) and incubated at 37°C for 2h. After incubation, the culture was centrifuged (4°C, 12000×g, 10min), and the supernatants were collected and filtered with a 0.22-μm nylon membrane before being subjected to the CE assay. The CE assay was conducted in a CE system, and the conditions to detect fructose were as follows: The detection wavelength was set at 254nm; the temperature was maintained at 25°C; and fused silica capillaries had dimensions of 50cm in total length and 40cm in effective length, and the inner diameter and outer diameter were 50μm and 375μm, respectively. The sample injection was maintained at an elevation difference of 20cm from the nearest detector for 5s, and the separation voltage was 25kV. The capillary system was rinsed with 0.1M NaOH for 5min, distilled water for 3min, and PBS for 3min before each run.

CE can also be used to detect the binding of drugs and proteins, which was used in this study to analyze the interaction of ursolic acid with GTF-SI and its variants as described previously (Liang et al., 2020). GTF-SI and its variants were dissolved in PBS (20mm, pH=6.8) and filtered with a 0.22-μm nylon membrane. Ursolic acid was dissolved in the same solution as that used to dissolve the proteins, and the final concentration was 0.3mg/ml. Afterward, the GTF or its variants were incubated with ursolic acid for 2h. Finally, CE was conducted as described above with a few differences: The detection wavelength was 210nm, and the separation voltage was 20kV.

Biofilms are the key factors in the induction of dental caries and periodontitis. We first measured the MIC and the MBC values of ursolic acid against S. mutans. The MIC value is the lowest concentration that could prevent the culture color from changing blue to pink. The MBC value is the lowest concentration that resulted in no visible bacterial colonies on the agar plates after incubation. As shown in Supplementary Figure S2 and shown in Table 1, both the MIC and MBC values of ursolic acid against S. mutans were 0.25mg/ml. Based on the concentration of antimicrobial activities, we evaluated the effect of ursolic acid on S. mutans biofilms by the XTT reduction method. As shown in Figure 1A, the bacterial viability within the biofilms decreased as the concentration of ursolic acid increased, while the difference was most significant at the concentration of 0.063mg/ml. Compared with the positive control, ursolic acid showed similar inhibitory effects on the bacterial viability of biofilms at 0.125mg/ml (Supplementary Figure S3). In addition, the crystal violet staining results showed that the amount of biofilm was decreased upon ursolic acid treatment, as shown in Figure 1B, which is consistent with the data of the XTT reduction assay.

To observe the direct effects of ursolic acid on biofilms, the Live/Dead BacLight Bacterial Viability kit was used to examine S. mutans bacteria within the biofilm, as shown in Figure 1. Compared with the control sample, the sample treated with ursolic acid had a large number of red-stained bacterial cells, indicating dead cells. Furthermore, the biofilms were thin. This result confirmed that ursolic acid could significantly destroy biofilms by affecting bacterial survival and adhesion.

EPSs are the key factors contributing to bacterial colonization, biofilm formation, maturation, and caries. Because EPSs are synthesized by GTFs, we explored the effect of ursolic acid on GTF activity in crude bacterial extracts in the presence of sucrose in vitro. As shown in Figure 2, ursolic acid inhibited the activity of GTFs, resulting in a decrease in the synthesis of EPS products. The inhibition of ursolic acid on water-soluble glucan production was stronger than that on water-insoluble glucan production at low concentrations of less than 0.06mg/ml, while this inhibition tendency was reversed at concentrations from 0.06mg/ml to 0.12mg/ml. Combined with the data above, these results suggest that ursolic acid could suppress the aggregation or adhesion of S. mutans to form cariogenic biofilms by preventing GTFs from synthesizing EPSs.

The catalytic center of GTF-SI has been shown to include two subsites for EPS synthesis: Subsite+1 contains the key amino acids for sucrose to bind, and subsite-1 contains the key amino acids for glucosyl moiety polymerization to form EPSs (Ito et al., 2011). In particular, Trp517 provides a platform for the acceptor glycosyl moiety of sucrose, and Tyr430 participates in hydrophobic interactions with carbon atoms of the glycosyl moiety in subsite+1; however, Asp909 and Tyr916 are related to recognition of the glucosyl moiety of the primary sucrose and formation of the glycosyl-enzyme intermediate in subsite-1 (Ito et al., 2011). As shown in Figure 3A, the simulation data showed that ursolic acid formed hydrogen bonds with Tyr430 and Asp909 of GTF-SI and had hydrophobic interactions with Leu433, Leu434, Phe907, Trp517, and Tyr916, which are the key sites for catalyzing sucrose to synthesize EPSs. Furthermore, the MD simulation results showed that the binding free energy of GTF-SI to ursolic acid was similar to that of sucrose (Supplementary Table S2). Then, we made two mutant GTF-SI variants in which the predicted amino acids that are essential for ursolic acid binding were replaced with other amino acids (Supplementary Table S2) to further verify the binding indicated by the simulation. The simulation showed that the number of amino acids in either GTF-SI variant that formed hydrogen bonds and hydrophobic interactions with ursolic acid decreased compared with that in wild-type GTF-SI (Figures 3B,C), and the interaction and binding free energy of the GTF-SI variants with ursolic acid were also decreased (Supplementary Table S2, S3).

Then, CE technology was used to further determine the interaction between GTFs and ursolic acid in vitro. As shown in Figure 3D, compared with that of the control, the migration time of the protein was prolonged by the addition of ursolic acid, which indicated that ursolic acid could bind with GTF-SI to change the charge-mass ratio, leading to a migration time-shift. The CE results further showed that the retention time of the variant A or variant B proteins was shorter than that of wild-type GTF-SI after the addition of ursolic acid, suggesting that the interaction between GTF-SI and ursolic acid became weaker or even lost as several key amino acids at the binding site were replaced (Figures 3E,F). These results indicate that ursolic acid might inhibit the enzymatic activity of GTFs through direct binding and that Phe907, Asp909 and Tyr916 in GTF-SI might be the key amino acids responsible for ursolic acid binding.

The process of GTF synthesis of EPS includes two steps: First, the GTF enzyme binds with sucrose, and then, EPS synthesis is catalyzed. After sucrose binds with subsite+1 of the GTF, the glucosyl group of sucrose dissociates from subsite+1 and then binds subsite-1 of the GTF to form an intermediate that catalyzes EPS production, while the fructosyl group of sucrose dissociates from subsite+1 of the GTF to produce the by-product, namely, fructose (Ito et al., 2011). Therefore, the production of fructose and EPSs would decrease when the binding site and catalytic site of the GTF are occupied by a molecule other than sucrose. Since the amino acids in GTF-SI that are essential for ursolic acid binding are also indispensable for sucrose recognition and since the MD simulation results showed that the binding free energy of GTF-SI to ursolic acid was similar to that of sucrose (Supplementary Table S2), we hypothesized that ursolic acid may suppress GTF activity by competitively blocking sucrose binding. To test this hypothesis, we evaluated the effects of ursolic acid on the production of EPS and fructose. To confirm the effect of ursolic acid on EPS synthesis by S. mutans, water-soluble glucans and water-insoluble glucans were obtained from bacterial culture treated with or without ursolic acid for 16h, and the phenol/H₂SO₄ method was used to detect the content of both types of EPSs. As shown in Figure 4A, ursolic acid at concentrations of 0.008mg/ml to 0.125mg/ml inhibited the production of both water-soluble and water-insoluble glucans. Additionally, the amount of fructose produced by crude recombinant GTF-SI was further measured in the presence of sucrose and ursolic acid for 2h. As shown in Figure 4B, the amount of fructose production decreased with the addition of ursolic acid to the enzymatic reaction system, and the effect of a high concentration was more obvious than that of a low concentration. Based on this in combination with the data above, we proposed that ursolic acid could compete with sucrose to occupy the catalytic center of the GTF, which then inhibits the enzymes that synthesize EPSs and eventually prevents bacterial adhesion to form biofilms.