Introduction

Dental caries remains a pressing global health challenge, with approximately 2 billion untreated permanent caries and 500 million deciduous caries cases reported globally in 20191. This pathology arises from dysbiosis of oral microbiota mediated by intricate host-diet-microbe interactions2. Frequent sugar intake induces rapid pH depression in biofilms3, shifting the demineralization-remineralization equilibrium toward irreversible enamel loss4, manifesting as white spots, cavitation, and eventual tooth destruction. Cariogenic biofilm formation follows a four-stage progression5: 1) Pioneer microbial colonization on enamel; 2) Exopolysaccharide (EPS)-mediated biofilm maturation; 3) Acidogenic/aciduric pathogen dominance establishing acidic niches; 4) Ecological collapse favoring acid-tolerant species. Streptococcus mutans (S. mutans), a keystone pathogen enriched in childhood caries6, outperforms commensals through EPS synthesis, adhesion, acidogenicity, and aciduricity7. While reducing fermentable carbohydrates effectively mitigates caries8, modern dietary habits render this impractical5.

Current interventions, including mechanical removal and chemical suppression (e.g., chlorhexidine), face limitations, such as patient compliance barriers and side effects9. Probiotics recalibrate oral ecology by depleting S. mutans in plaque and saliva10, yet raise safety concerns due to viability challenges and potential cariogenic risks5. Postbiotics emerge as safer alternatives. The ISAPP defined postbiotics as “preparation of inanimate microorganisms and/or their components that confer a health benefit on the host”11. Postbiotics exhibit superior pH/thermal stability, storage tolerance, and clinical safety compared to live probiotics11,12. Common forms include heat-killed probiotics and cell-free supernatant (CFS). Within China’s probiotic oral hygiene sector, Lacticaseibacillus paracasei (L. paracasei) is a recognized and well-adopted ingredient. L. paracasei demonstrates exceptional anticaries potential. The postbiotic lozenges (including Ligilactobacillus salivarius subsp. salicinius AP-32, L. paracasei ET-66, and Lactiplantibacillus plantarum LPL28) can boost the concentration of salivary IgA, lead to a significant decline in the number of S. mutans, and raise the population of beneficial oral bacteria13. However, despite promising in vivo outcomes, mechanistic insights into its postbiotic-mediated suppression of S. mutans remain underexplored.

To end this, this study investigates the effects of L. paracasei postbiotics on planktonic and biofilm S. mutans. It utilized in vitro models to evaluate the ability of postbiotics to suppress the release of free calcium. Finally, it employs multi-omics approaches to further clarify the inhibitory mechanisms of postbiotics against S. mutans. Specifically, we (i) evaluated the antibacterial and anti-biofilm effects of CFS derived from L. paracasei against S. mutans by assessing membrane integrity, cell viability, and surface hydrophobicity, (ii) characterized the biofilm disruption by CFS, including changes in biofilm biomass, metabolic activity, bacterial chain length, and EPS accumulation through SEM and CLSM imaging, (iii) established an in vitro hydroxyapatite model mimicking the oral environment to simulate the impact of CFS on biofilm formation and enamel demineralization, and (iv) used integrated transcriptomics and metabolomics to elucidate the molecular mechanisms, identifying key genes and metabolites associated with biofilm regulation, virulence, and quorum sensing pathways.

Results and Discussion

The Effects of CFS on Planktonic Streptococcus mutans

Compared to the control, CFS exhibited significant inhibitory effects at final concentrations of 1/2×, 3/4×, 1×, and 3/2×, with no notable differences observed between the latter three concentrations (Fig. 1A). Given that the 1× CFS accurately reflects the bacteriostatic effect at the original concentration of inhibitory substances in the fermented broth, this dosage was selected for subsequent planktonic cell experiments.

Fig. 1: Effects of CFS on planktonic bacterial cells.
Fig. 1: Effects of CFS on planktonic bacterial cells.
Full size image

A Inhibitory effects of CFS at different final concentrations on S. mutans (Tukey test). B Impact of CFS on the growth curve of S. mutans. C Effect of CFS treatment on cell surface hydrophobicity of S. mutans (t test). D Morphological changes of S. mutans treated with CFS (SEM images): a1 and a2, control group (5000× and 20000× magnification); b1 and b2, CFS-treated group (5000× and 20000× magnification). White rectangles indicate the fields selected for 20000× magnification. E Histogram of fluorescence intensity for S. mutans. Mean fluorescence intensity (MFI) of F CFDA, G DiBAC4(3), and H PI (t test). *** P < 0.001. Different letters indicate statistically significant differences (P < 0.05).

Growth profiles were fitted using the SRichards2 model, yielding an R2 of 0.99967 for the control group. CFS treatment completely suppressed S. mutans proliferation, with no observable growth phase (Fig. 1B).

SEM imaging revealed stark contrasts between groups. Control cells displayed smooth, intact surfaces (Fig. 1D-a1/a2), whereas CFS-treated cells exhibited severe membrane wrinkling (red arrows), surface roughness, and extracellular debris (Fig. 1D-b1/b2). These findings align with Nataraj et al.14, who reported similar membrane damage (e.g., pore formation, cytoplasmic leakage) in Staphylococcus aureus treated with biosurfactants from Lactobacillus acidophilus NCFM and Lacticaseibacillus rhamnosus GG.

SEM observations suggested that CFS increases membrane permeability, leading to cytoplasmic leakage, which may contribute to bacterial viability loss. To validate this hypothesis, PI was used to label cells with permeabilized membranes, and CFDA was used to selectively stain metabolically active cells. Notably, membrane integrity disruption often induces membrane depolarization15. A normal membrane potential (ΔΨ) is essential for bacterial virulence and antibiotic tolerance16. For instance, melittin B17 disrupts lipid bilayers, thereby collapsing ΔΨ. The collapse of the ΔΨ could disrupts critical bacterial processes. To probe ΔΨ changes, DiBAC4(3) was employed, which fluoresces upon binding depolarized membranes18. CFS treatment drastically reduced CFDA+ cell counts and MFI (Fig. 1E, F), confirming viability loss. Concurrently, DiBAC4(3)+ (Fig. 1E, G) and PI+ (Fig. 1E, H) populations increased, demonstrating ΔΨ collapse and permeability enhancement. These results correlate with SEM-based morphological evidence.

CFS reduced surface hydrophobicity from 93.66% ± 1.46% (control) to 69.81% ± 3.69% (Fig. 1C). Hydrophobicity governs critical physiological activities of S. mutans, including adhesion, aggregation, and biofilm formation19,20. This hydrophobicity is mediated by hydrophobic side chains or terminal groups present on the cell surface21. The morphological disruption observed in Fig. 1D suggests that CFS may induce alterations in these hydrophobic components, thereby reducing the hydrophobicity of S. mutans.

The Effects of CFS on Biofilms

CFS treatment reduced S. mutans biofilm biomass by 17.0% ± 3.4%, while 2× concentrated CFS (2CFS) achieved a 27.6% ± 3.5% reduction (Fig. 2A). These results demonstrate CFS’s capacity to disrupt biofilm structural integrity, despite the inherent resilience of biofilm-embedded cells to antimicrobial agents.

Fig. 2: Effects of CFS on biofilms.
Fig. 2: Effects of CFS on biofilms.
Full size image

A Biomass quantification of biofilms by crystal violet assay (Tukey test). B Proliferation capacity of biofilm-embedded S. mutans assessed via CCK-8 assay (t test). C Impact of CFS on bacterial chain length within biofilms. When three or more cocci are arranged in a linear configuration, they are considered a chain for length measurement purposes. D Effects of CFS treatment on the surface morphology of S. mutans biofilms (SEM Images). E Composition and 3D structure of S. mutans biofilms analyzed by CLSM. EPS, viable cells, and cells with compromised membrane integrity were stained with Rhodamine-ConA (orange), CFDA (green), and PI (red), respectively. Scale bars: 50 μm (white bars in E), 20 μm (grids in 3D view). *** P < 0.001. Different letters indicate statistically significant differences (P < 0.05).

Bacterial proliferation within biofilms is a critical factor in S. mutans pathogenicity due to the enhanced tolerance of biofilm-embedded cells. The CCK-8 assay revealed >99% inhibition of bacterial proliferation in CFS-treated biofilms (Fig. 2B). OD450 decline after 6 h may reflect acid metabolite accumulation during late-stage S. mutans growth, potentially interfering with chromogenic reactions.

SEM imaging showed that CFS-treated biofilms were thinner, flatter, and sparsely populated compared to controls (Fig. 2D-a1/b1). Analysis of three random fields at 1000× magnification per group revealed the chain length distribution (Fig. 2C). The control group had a chain length distribution with a mode of 4 and a maximum of 10, whereas the CFS group had a distribution with a mode of 3 and a maximum of 6. This aligns with the study of Tahmourespour et al.22, which reported similar chain shortening in S. mutans biofilms exposed to Lactobacillus-derived biosurfactants.

These findings from CLSM experiments collectively indicate that CFS disrupts biofilm ecology (Fig. 2E) via viability suppression (CFDA+ depletion), membrane damage (PI+ accumulation), and EPS degradation (Rhodamine-ConA signal reduction).

This study prioritized high-throughput and in-situ analysis of live bacteria in the biofilm using CCK-8 and CLSM, in favor of direct CFU counts—a recognized methodological limitation.

Bioactive Components and Application Potential of CFS

Probiotic-derived antimicrobials primarily include bacteriocins, enzymes, organic acids, biosurfactants, hydrogen peroxide, and EPS23. Improve the inactivation strategy proposed by Wasfi et al.24, and investigate the main antibacterial substances of the CFS. Neither protein denaturation (via neutral protease) nor H₂O₂ elimination (via catalase) diminished antibacterial activity, ruling out major roles for proteins and hydrogen peroxide (Fig. 3A). Supplemented EPS showed negligible effects, likely due to low concentration (Fig. 3B). Moreover, the protease resistance and thermal stability of CFS indicated that biosurfactants—primarily composed of glycoproteins and glycolipopeptides in L. paracasei25—were not primary contributors.

Fig. 3: Bioactive Components and Application Simulation Analysis of CFS.
Fig. 3: Bioactive Components and Application Simulation Analysis of CFS.
Full size image

A Antimicrobial activity after inactivation of potential bioactive components (Tukey test). CAT denotes catalase. B Antimicrobial activity of EPS extracted from CFS (Tukey test). C pH stability of CFS. D Effects of HCl/NaOH-simulated acidic/alkaline environments on S. mutans (Tukey test). E Thermal stability of CFS (Tukey test). F-H Storage stability of CFS under different temperatures after 1 week (F), 1 month (G), and 3 months (H) (Tukey test). I Side and top views of S. mutans oral biofilm in the in vitro model. J Dry weight of biofilms in the in vitro model. K pH of spent media from the in vitro model. L Release of free calcium from HA disks in the in vitro model. As shown in Fig. 3I–L, the negative control (NC), 2CFS, and positive control (PC) groups in the in vitro model were treated with BHI, 2CFS, and 0.01% CHX as oral rinse simulation fluids, respectively. For experiments employing HA disks, one disc was used per condition, with three independent biological replicates performed for each. Different letters indicate statistically significant differences (P < 0.05). ns, no significant difference.

By contrast, neutralizing acidic components (pH adjusted to 6.50) markedly reduced CFS inhibition (Fig. 3A), strongly implicating acids as critical agents. When inorganic acid was used to simulate an acidic environment, it was found that acid stress alone failed to significantly inhibit the proliferation of S. mutans (Fig. 3D). In the in vitro model of S. mutans biofilm, the pH of the culture medium dropped below 4 after 14 hours of incubation (Fig. 3K). This is consistent with the fact that S. mutans is recognized as a cariogenic pathogen largely due to its notable acidogenicity and acid tolerance. Taken together, these results indicate that the antibacterial activity of L. paracasei CFS (pH 4) is not solely attributable to low pH, but likely involves the specific synergistic effects of various organic acid. Similarly, Lim et al. reported that the inhibitory effect of Weissella cibaria on periodontal pathogens was dependent on both acid and hydrogen peroxide concentrations26.

The pH of commercially available mouthwashes typically ranges from 4 to 827, which likely represents the pH environment during the industrial production of CFS. When the pH of CFS was adjusted to 2 and 4, the OD600 showed no statistically significant differences compared to untreated CFS (Fig. 3C). However, at pH 6 and 8, an increase in OD600 values was observed (Fig. 3C). Therefore, careful selection of appropriate pH conditions is critical for practical applications to balance industrial feasibility and biological effectiveness. Acid/base controls (HCl/NaOH-adjusted water) demonstrated no significant inhibitory effects (Fig. 3D). No activity loss was observed after heat treatment (40 °C–121 °C; Fig. 3E), supporting CFS’s suitability for industrial processing. CFS maintained activity across temperatures of -20 °C to 37 °C when stored for durations ranging from 1 week to 3 months.

HA disks (mimicking enamel28) coated with human saliva were subjected to 7-day simulated oral cycles (feeding/rinsing). Compared to negative controls (BHI), 2CFS-treated biofilms displayed a thinner and smoother morphology with fewer protrusions (Fig. 3I), which was consistent with the SEM imaging (Fig. 2D). While 2CFS reduced biomass compared to the negative controls (Fig. 3J), its efficacy was still slightly lower than that of CHX, the clinical gold standard29 for oral antimicrobial therapy. Despite CHX’s documented side effects, such as taste distortion and tooth staining30, it remains the preferred choice in many clinical settings. Although the extracellular pH of 2CFS (4.00 ± 0.02) was comparable to that of the controls (3.95 ± 0.01) (Fig. 3K), localized acid accumulation was the primary driver of HA demineralization. Moreover, the release of free calcium from HA serves as a key indicator for assessing cariogenic potential. CFS significantly reduced the release of free calcium (Fig. 3L), pointing to its potential to effectively reduce the cariogenicity of S. mutans.

Mechanistic Insights into CFS-Mediated Suppression of Streptococcus mutans

Using transcriptomics and metabolomics, this study investigates the effects of CFS on virulence factors and QS-related genes and metabolites in S. mutans. Key enriched pathways (Gene Ratio > 0.4) included “Biofilm formation” and “Starch and sucrose metabolism” (Fig. 4A), both tightly linked to EPS synthesis and adhesion. Differential metabolites were predominantly enriched in KEGG “Metabolism” pathways across positive/negative ion modes (Fig. 4D).

Fig. 4: KEGG pathway enrichment and differential analysis of DEGs and Metabolites in S. mutans.
Fig. 4: KEGG pathway enrichment and differential analysis of DEGs and Metabolites in S. mutans.
Full size image

A KEGG pathway enrichment bubble plot of DEGs. Bubble size indicates the number of DEGs per pathway; color gradient reflects enrichment significance (Q-value: adjusted P-value via BH method; Gene Ratio: number of enriched DEGs / total genes in pathway). B Fold change of DEGs associated with virulence factors and QS in S. mutans. C Relative expression of virulence- and QS-related genes in S. mutans (Dunnett test). D KEGG pathway enrichment bar plot (level 2 classification) of differential metabolites. ns, no significant difference, * P < 0.05, ** P < 0.01.

Genes related to virulence factors and QS were screened and their relative expression levels were displayed (Fig. 4B). The synthesis of EPS is primarily mediated by Glycosyltransferases (Gtfs), including gtfA, gtfB, and gtfC31,32. CFS treatment upregulated the expression of gtfA, gtfB, and gtfC (Fig. 4B), while significantly downregulating stsR—a transcriptional regulator critical for sugar transport, EPS production, and biofilm formation33. The suppression of stsR likely disrupted EPS assembly, thereby explaining the observed reduction in biofilm polysaccharides. Glucan-Binding Proteins (Gbps) facilitate sucrose-dependent adhesion by binding glucans synthesized via Gtfs34. Specifically, gbpA promotes glucan-dependent biofilm maturation35. gbpB governs sucrose-dependent biofilm integrity, hydrophobicity, and cell wall maintenance36. gbpC mediates glucan-dependent bacterial aggregation35. Under CFS treatment, gbpA and gbpB were downregulated, whereas gbpC was upregulated (Fig. 4B). These alterations may collectively impair biofilm cohesion and reduce hydrophobicity. eno (enolase), responsible for phosphoenolpyruvate (PEP) synthesis37, was upregulated. ldh (lactate dehydrogenase), pivotal for acid production during carbohydrate metabolism38, and scrB (sucrose-6-phosphate hydrolase), which cleaves internalized sucrose-6-phosphate into glucose-6-phosphate and fructose39, were both downregulated (Fig. 4B). This metabolic shift likely contributed to decreased acid production, consistent with the observed reduction in free calcium release in the in vitro model. The agmatine deiminase system (AgDS), which neutralizes cytoplasmic pH for acid resistance40, relies on aguA (agmatine deiminase) and aguB (putrescine carbamoyltransferase)41. CFS suppressed aguB but upregulated aguA (Fig. 4B). F0F1-ATPase, a proton pump regulating cytoplasmic pH, is encoded by atpA (α subunit), atpD (β subunit), atpB (A subunit), and atpF (B subunit)42. atpD is a key determinant of S. mutans aciduricity31; its downregulation, alongside atpA, likely impaired acid adaptation, while atpB and atpF upregulation suggests compensatory mechanisms (Fig. 4B).

The QS system regulates virulence and biofilm formation by releasing, sensing, and interacting with diffusible signaling molecules in response to cellular density within the surrounding environment34,43. In Gram-positive bacteria, QS typically relies on signaling peptides, histidine kinase (HK) sensors, and response regulators (RRs)44. The latter two form a two-component signal transduction system (TCSTS), a critical regulatory network for bacterial adaptation, survival, and virulence45. The LuxS-mediated QS system facilitates biofilm formation and enhances stress resistance in response to environmental challenges46. vicRK signal transduction is closely related to sucrose dependent adhesion and biofilm formation of S. mutans. vicR knockout is lethal45, while its overexpression promotes abnormal biofilm clustering and elongated bacterial chains47. The ComD receptor (encoded by comD) interacts with competence-stimulating peptide (CSP, encoded by comC) to drive biofilm development48. ciaRH operon is related to biofilm formation, acid tolerance, and bacteriocin production49. Calcium signaling modulates this operon via ciaX, which mediates calcium-dependent biofilm dynamics49,50. ciaR mutation reduces polysaccharide synthesis, destabilizing biofilms51, while ciaH inactivation in S. mutans UA140 suppresses bacteriocin production, sucrose-dependent biofilm formation, and acid sensitivity50. CFS downregulated the luxS, ciaR, ciaH, and ciaX (Fig. 4B), indicating that the mechanism may involve reducing cariogenicity by suppressing the LuxS/AI-2 system and the ciaRHX operon. However, CFS upregulation of comD, vicK, and vicR genes was observed, which may be analogous to L. salivarius CFS, where acid/H2O2 stressors upregulate comC/comD, triggering CSP-mediated countermeasures24.

qPCR results (Fig. 4C) showed that the upregulation and downregulation trends of the gtfB, stsR, gbpB, scrB, aguB, atpA, luxS, and ciaR were consistent with those in the transcriptomic data. The alignment between the qPCR results and omics data reinforces the reliability of our findings. It suggests that the regulatory effects of L. paracasei postbiotics on these genes are significant.

Among the pathways annotated by 23 genes associated with virulence factors and QS (Fig. 4B), 37 metabolites that were also annotated by metabolomics in this study were screened out. A correlation heatmap revealed significant associations between these genes and metabolites (Fig. 5A), while a connectivity network highlighted strongly correlated pairs (|r | > 0.8; Fig. 5B). Creatine was the highest connectivity in the network (Fig. 5B). Creatine has been detected in early childhood caries biofilms52 but has not been functionally characterized in bacterial contexts. In vertebrates, creatine modulates immune responses and glucose metabolism53. Here, CFS treatment reduced creatine abundance, which positively correlated with scrB, stsR, and ciaX, but negatively correlated with gtfC, gbpC, atpF, vicK, and comD (Fig. 5B). PEP, a critical component of the carbohydrate phosphotransferase system (PTS)37, was the second most highly connected in the network. Strong negative correlation (r = -0.855) with ldh expression. The downregulation of ldh induced by CFS may have resulted in PEP accumulation, thereby disrupting acid production—a change concordant with the observed reduction in free calcium release in the in vitro model.

Fig. 5: Integrated Analysis of Transcriptomics and Metabolomics.
Fig. 5: Integrated Analysis of Transcriptomics and Metabolomics.
Full size image

A Correlation heatmap between virulence-associated gene expression and differential metabolite abundance. Note: Circles represent correlation strength, with circle size proportional to the absolute value of Pearson’s correlation coefficient (|r | ). The r value indicates Pearson’s correlation between the expression of virulence factors/QS-related genes and metabolite abundance; positive r (red) denotes positive correlation, negative r (blue) denotes negative correlation. B Connectivity network of strongly correlated gene-metabolite pairs (|r | > 0.8). Nodes (genes, triangles; metabolites, stars) are colored by connectedness. Edges represent correlation strength and sign, using solid (positive) or dashed (negative) lines. * P < 0.05, ** P < 0.01, ***P <0.001.

Materials and methods

Strain

S. mutans CGMCC 1.2499 and L. paracasei CGMCC 1.2744 were obtained from the China General Microbiological Culture Collection Center. All strains were preserved using standard microbiological methods prior to use. Strains were cultured in BHI broth (OXOID) and MRS broth (Huankai, China; pH 6.80 ± 0.05), respectively.

Preparation of L. paracasei CFS

The fermented culture, prepared by inoculating 5% seed liquid and culturing at 37 °C for 24 h, was centrifuged (4500 × g, 10 min, 4 °C) to collect the supernatant (pH 4.00 ± 0.20), which was filter-sterilized through a 0.22 μm membrane to obtain CFS. The CFS was concentrated to 50% of the original volume using a rotary evaporator (60 °C, 180 rpm), yielding 2CFS.

Validation System

A modified liquid culture system24 was employed. Activated culture, treatment solution, and 2× BHI broth were mixed at a ratio of 1:4:5 (v/v) and incubated at 37°C for 12 h. The OD600 of each group was measured using a Synergy H1MF microplate reader (Agilent). In assays utilizing planktonic S. mutans, sterile ultrapure water and 2CFS were employed for the control and experimental groups, respectively. The use of sterile ultrapure water—rather than the MRS medium, which showed no significant growth inhibition (data not shown)—as the control ensured the exclusion of potential interference from other components. In the validation system, replacing the activated bacterial suspension with BHI gave the blank sample group.

The Effects of CFS on Planktonic Streptococcus mutans

OD600 was measured hourly during 0–11 h of incubation. Treated cultures were centrifuged (4500 ×g, 5 min), resuspended in 5 mM PIPES buffer (pH 6.5) containing 2.5% glutaraldehyde, and fixed overnight at 4°C. After centrifugation, pellets were sequentially dehydrated with 30, 60, 90%, and absolute ethanol (10 min each). Samples were imaged using SEM (Zeiss Gemini300).

Centrifuge the bacterial suspension at 4 °C for 3 min and discard the supernatant. Dual staining was performed using 10 mM CFDA (green fluorescence, live cells), 5 μM DiBAC4(3) (green fluorescence, membrane depolarization), and 75 mM PI (red fluorescence, dead cells). After 15 min incubation at 37°C and PBS washing, samples were filtered through 300-mesh nylon (Solarbio) and analyzed via flow cytometry (BD Accuri C6 Plus).

Three milliliters of cell suspensions (OD660 = 0.55 ± 0.05, denoted as A0) were mixed with 400 μL hexadecane (Vokai), vortexed for 90 s, and settled for 15 min. Hydrophobicity was calculated as (A0 − A1)/ A0 × 100%, where A1 represents the aqueous phase absorbance. PUM buffer served as a blank control.

The Effects of CFS on Biofilms

To support and maintain biofilm structural integrity, BHI —rather than the sterile ultrapure water used for planktonic cells—was employed as the control in S. mutans biofilm experiments.

S. mutans inoculum was diluted 10-fold in BHI broth supplemented with 2% sucrose (w/v) (BHI-2%s) to an OD600 of 0.20 ± 0.05, then incubated statically at 37 °C for 12 h. Discard the liquid and wash the biofilm once with fresh BHI broth.

Preformed biofilms in 96-well plates were treated with 100 μL CFS/2CFS (experimental) or BHI (control) for 24 h at 37°C. After PBS washing, biofilms were fixed with methanol (15 min), stained with 0.017% ammonium oxalate crystal violet (Solarbio, China), and dissolved in ethanol. Absorbance (OD570) was measured. Inhibition rate = (Control OD570 − Experimental OD570)/Control OD570× 100%.

Preformed biofilms in 96-well plates were treated with 70 μL CFS + 35 μL BHI (experimental) or 105 μL BHI (control). CCK-8 kit (10 μL, MCE) was added, and OD450 was measured at 2, 4, 6, and 8 h. Inhibition rate = [(Control OD450−Blank) − (Experimental OD450−Blank)] / (Control OD450−Blank) × 100%.

Preformed biofilms on coverslips were treated with 4 mL BHI (control) or 2CFS (experimental) for 24 h at 37°C. Fixation and dehydration followed the procedures described in the section of “The Effects of CFS on Planktonic Streptococcus mutans” for SEM sample preparation, omitting centrifugation during solution replacement.

Preformed biofilms in confocal dishes were treated with 1 mL BHI (control) or 2CFS (experimental) for 24 h. After PBS washing, dual staining was performed using 60 μM PI and 100 μg/mL Rhodamine-ConA (orange fluorescence, EPS) with 100 μM CFDA. Z-stack images (0.5 μm intervals) were acquired via CLSM (Zeiss LSM800, 60× oil immersion).

Bioactive Components and Application Simulation of CFS

We explored the major class of active substances in CFS by inactivating them and extracting crude active substances. ① H2O2 neutralization: Excess catalase (10 U/mL; Aladdin) was added to 2CFS (containing 3.5 μM H2O2, data not shown), incubated at 37 °C for 2 h, and heat-inactivated (98 °C, 5 min). ② Protein degradation: 2CFS was treated with 100 U/mL neutral protease (Solarbio) at 37 °C for 2 h, followed by heat inactivation. ③ pH adjustment: CFS was neutralized to pH 6.50 ± 0.05 (simulating salivary pH) using NaOH. ④ Acid control: Ultrapure water was acidified to pH 4.00 ± 0.05 (matching native CFS) with HCl. ⑤ EPS extraction: Crude EPS was precipitated from 2CFS by adding three volumes of ethanol, overnight incubation, centrifugation (4500 × g, 5 min), and drying. EPS dosage was calculated as: (Dry weight of EPS (mEPS) / Volume of 2CFS used for extraction) × Volume of 2CFS in validation system. Validation systems included control groups (sterile water or acidified water, pH 4.00) and experimental groups (2CFS treated as described above and EPS).

2CFS and ultrapure water were adjusted to pH 2, 4, 6, 8, 10 ( ± 0.05) using HCl/NaOH, filter-sterilized, and tested in validation systems. Controls included sterile water, untreated 2CFS, and pH-adjusted water.

2CFS was heated at 40 °C, 60 °C, 80 °C, 100 °C (1 h) or autoclaved (121 °C, 15 min). Treated samples were evaluated in validation systems against untreated 2CFS and sterile water controls.

Aliquots of 2CFS were stored at -20 °C, 4 °C, 25 °C, and 37 °C in light-proof tubes. Stability was assessed at 1 week, 1 month, and 3 months using validation systems with sterile water as the negative control.

In Vitro Biofilm Analysis was preformed followed below steps: ① biofilm formation was performed using a modified preparation method54 (see Fig. 6). Hydroxyapatite disks (HA disks; QBioscience) were incubated with human saliva (37 °C, 20 min). S. mutans inoculum was diluted 10-fold in BHI supplemented with 4% sucrose (w/v) (BHI-4%s), and 1 mL was added to saliva-coated HA disks in 24-well plates. After 24 h static incubation at 37°C, the medium was replaced with fresh BHI-4%s for an additional 24 h. ②Simulation experiment: Simulated daily oral procedures including feeding and mouth rinsing were conducted at 7:00, 12:00, and 17:00 for 7 consecutive days. The negative control (NC), experimental, and positive control (PC) groups used BHI broth, 2CFS, and 0.01% chlorhexidine (CHX, Yuanye) as their respective simulated mouthrinse solutions. A. Feeding simulation: Medium was replaced with 1 mL BHI-4%s and incubated statically (37 °C, 20 min). B. Rinsing simulation: One milliliter of simulated rinsing solution was added and shaken at 200 rpm for 5 min. After removal, biofilms were washed once with sterile PBS. C. Residual simulation: 5 mL BHI-2%s was added to mimic food debris. ③ Biofilm biomass: HA disks were photographed (lateral and top views), boiled in 2 mL PBS (10 min) to detach biofilms, and centrifuged (7425 × g, 10 min). Pellets were washed once with PBS, dried at 45°C, and weighed. ④ pH monitoring: Spent morning media were collected daily for pH measurement. ⑤ Free calcium analysis: All spent fluids (total volume V) were centrifuged (7425 × g, 10 min), filter-sterilized, and analyzed via ICP-OES (Agilent 720ES). Baseline solutions were prepared by mixing BHI-2%s, BHI-4%s, simulated rinsing solutions, and PBS (1:1:1:1). The amount of free calcium released was calculated as: Δm = (C₀ − Cbaseline) × V / 1000 (mg), where C₀ = calcium concentration in test samples, Cbaseline = baseline calcium concentration.

Fig. 6: Preparation process of the in vitro biofilm model.
Fig. 6: Preparation process of the in vitro biofilm model.
Full size image

The dashed horizontal line represents the timeline, indicating 7 a.m., 12 p.m, and 5 p.m. from top to bottom. Vertical lines denote experimental days. Circles at the intersections indicate specific operations on corresponding days: light yellow for HA discs incubated with saliva at 37°C for 20 min; dark yellow for replacement of the solution with BHI-4%s; and red for replacement with BHI-4%s incubated at 37°C for 20 min, followed by treatment with the mouthwash mimic under shaking at 120 rpm for 5 min, washing with PBS, and final replacement with BHI-2%s.

Mechanistic Exploration of CFS-Mediated Streptococcus mutans Inhibition

Preformed biofilms (prepared as described in the section of “Bioactive Components and Application Simulation of CFS” for in vitro biofilm analysis) were collected, centrifuged (4500 ×g, 4°C, 3 min), flash-frozen in liquid nitrogen, and shipped on dry ice to GeneDenovo Co. for sequencing.

Total RNA was extracted using TRIzol reagent (Life Technologies, USA). RNA quality was assessed by 1% agarose gel electrophoresis and quantified using a NanoDrop 2000 spectrophotometer (Thermo Scientific). Precise quantification and integrity assessment were performed on an Agilent 4200 TapeStation, confirming all samples had an RNA Integrity Number (RIN) > 8.0 and a total mass > 0.5 μg. Ribosomal RNA was subsequently depleted from 1 μg of qualified total RNA using the Ribo-Zero rRNA Removal Kit (Bacteria) (Illumina). The second-strand cDNA was degraded with the USER Enzyme (New England BioLabs) and purified using Agencourt AMPure XP beads (Beckman Colter), followed by cluster generation on a cBot system with NEBNext Q5 Hot Start HiFi PCR Master Mix (New England BioLabs). The libraries were sequenced on an Illumina NovaSeq 6000 platform for 150 bp paired-end reads. Raw reads were filtered using FASTP to obtain clean data (Table S1). Ribosomal RNA reads were removed via Bowtie2 alignment (Table S2). Intra-group Pearson correlation coefficients and principal component analysis (PCA) were calculated using R packages psych and gmodels, respectively (Fig. S1A). Differentially expressed genes (DEGs) were identified with DESeq2 (FDR < 0.05, |FoldChange | > 1.2; Fig. S1B) and annotated to Gene Ontology (GO) terms (Molecular Function, Figure S1C; Biological Process, Fig. S2A) and KEGG pathways (Metabolism, Fig. S2B, Table S6).

DEGs associated with virulence factors and QS were screened, and more literature-reported genes were validated by qPCR with primers listed in Table S3. RNA extraction from biofilms followed the section of “Mechanistic Exploration of CFS-Mediated Streptococcus mutans Inhibition” for transcriptomic analysis, using the FastPure Cell/Tissue Total RNA Isolation Kit V2 (Vazyme). cDNA was synthesized with TransScript® All-in-One First-Strand cDNA Synthesis SuperMix for qPCR (TransGen Biotech). qPCR reactions (Table S4) were performed with THUNDERBIRD SYBR Mix (Toyobo) on a qTOWER3G system (Analytik Jena AG) under cycling conditions in Table S5. Relative gene expression was calculated via 2-∆∆Ct method: -∆∆Ct = (Ct target – Ct reference) control - (Ct target - Ct reference) experimental.

Samples were derived from the same batch as those used for transcriptomic analysis. Prior to flash-freezing in liquid nitrogen, biofilms were washed once with PBS and processed for LC/MS analysis. Metabolites were separated using a hydrophilic interaction liquid chromatography (HILIC) column on an Agilent 1290 Infinity LC UHPLC system. MS and MS/MS spectra were acquired in both positive and negative ion modes on an AB Sciex TripleTOF 6600 mass spectrometer. Compound annotation was performed using public databases (Metlin, MoNA, MassBank) and an in-house MS/MS library. Data quality was assessed via PCA and PLS-DA (Fig. S3A). Differential metabolites were defined by Variable Importance in Projection (VIP) ≥ 1 and univariate t-test P < 0.05, with significant enrichment in KEGG metabolic pathways (Fig. S3C) and GO terms (Figure S3D). Volcano plots visualized up-/downregulated metabolites (Fig. S3B).

Pearson correlation heatmaps (coefficients ranging −1 to +1) evaluated gene-metabolite relationships55. Strongly correlated pairs (|r | > 0.8) were mapped onto connectivity networks.

Statistical Analysis

Data are presented as mean ± SD, with specific details provided in figure legends. All experiments were performed with at least three independent biological replicates. Normality was assessed using the Shapiro-Wilk test. For normally distributed data, an independent samples t-test was used for two-group comparisons, and one-way ANOVA followed by Tukey’s post hoc test for multi-group comparisons. Non-normally distributed data were analyzed using Dunnett’s test.