Iron oxide nanozymes stabilize stannous fluoride for targeted biofilm killing and synergistic oral disease prevention

Dental caries (tooth decay) is the most prevalent human disease caused by oral biofilms, affecting nearly half of the global population despite increased use of fluoride, the mainstay anticaries (tooth-enamel protective) agent. Recently, an FDA-approved iron oxide nanozyme formulation (ferumoxytol, Fer) has been shown to disrupt caries-causing biofilms with high specificity via catalytic activation of hydrogen peroxide, but it is incapable of interfering with enamel acid demineralization. Here, we find notable synergy when Fer is combined with stannous fluoride (SnF2), markedly inhibiting both biofilm accumulation and enamel damage more effectively than either alone. Unexpectedly, our data show that SnF2 enhances the catalytic activity of Fer, significantly increasing reactive oxygen species (ROS) generation and antibiofilm activity. We discover that the stability of SnF2 (unstable in water) is markedly enhanced when mixed with Fer in aqueous solutions without any additives. Further analyses reveal that Sn2+ is bound by carboxylate groups in the carboxymethyl-dextran coating of Fer, thus stabilizing SnF2 and boosting the catalytic activity. Notably, Fer in combination with SnF2 is exceptionally effective in controlling dental caries in vivo, preventing enamel demineralization and cavitation altogether without adverse effects on the host tissues or causing changes in the oral microbiome diversity. The efficacy of SnF2 is also enhanced when combined with Fer, showing comparable therapeutic effects at four times lower fluoride concentration. Enamel ultrastructure examination shows that fluoride, iron, and tin are detected in the outer layers of the enamel forming a polyion-rich film, indicating co-delivery onto the tooth surface. Overall, our results reveal a unique therapeutic synergism using approved agents that target complementary biological and physicochemical traits, while providing facile SnF2 stabilization, to prevent a widespread oral disease more effectively with reduced fluoride exposure.


Introduction
Dental caries is the most prevalent and costly bio lm-induced oral disease that causes the destruction of the mineralized tooth tissue 1 . In caries-inducing (cariogenic) bio lms, microorganisms form highly protected biostructures that create acidic pH microenvironments, promoting cariogenic bacteria growth and acid dissolution of tooth-enamel 2, 3 . Despite increased use of uoride (the mainstay anticaries agent), it remains unresolved and affects 3.1 billion people worldwide, with annual costs exceeding US $290 billion 4,5 . Even though uoride is effective in reducing tooth enamel demineralization at acidic pH values 6, 7 , it has limited antibio lm activity despite inhibitory effects against planktonic bacteria 8 . Additionally, current modalities, including high-dose uoride treatments, are insu cient to prevent dental caries in high-risk individuals where pathogenic dental bio lms rapidly accumulate under sugar-rich diets and poor oral hygiene that enables rm bacterial adhesion to teeth 9,10 . Furthermore, the level of exposure to uoride that provides strong protection has accompanying risks (e.g., dental uorosis), especially for children 11,12,13 , as uoride overexposure has detrimental effects 14,15 .
Ferumoxytol (Fer), an aqueous iron oxide nanoparticle formulation approved by the Food and Drug Administration (FDA) for systemic treatment of iron de ciency, has shown both e cacy and speci city against cariogenic bio lms when used topically, through selective pathogen binding and acidic pHactivation of hydrogen peroxide (H 2 O 2 ) via catalytic (peroxidase-like) activity 16,17 . Although topical applications of Fer can reduce dental caries in vivo, it does not interfere physiochemically with enamel demineralization and is unable to entirely prevent the progression of the disease. To improve the e cacy of Fer, combination with uoride could potentiate the therapeutic effects. We hypothesized that Fer and uoride could complement each other's properties, even at lower concentrations, to target the development of dental caries more effectively without increasing uoride exposure.
Herein, we evaluated the combination of Fer with two formulations of uoride widely used in oral health care, sodium uoride (NaF) and stannous uoride (SnF 2 ). While the combination with NaF did not show improvement, when Fer was combined with SnF 2 we observed remarkable synergistic effects in vitro and in vivo. We found that SnF 2 was highly stable in aqueous solution when mixed with Fer; the lack of stability of SnF 2 has been a major limitation in commercial formulations, requiring use of chemical additives 18, 19 . Unexpectedly, the catalytic activity of Fer signi cantly increased when mixed with SnF 2 , thereby enhancing antimicrobial potency. Further analysis revealed that Sn 2+ was bound by carboxylate groups in the carboxymethyl-dextran coating of Fer, thereby enhancing the stability of SnF 2 . When tested in a rodent model, we found that Fer in combination with SnF 2 was exceptionally effective in preventing dental caries (substantially superior to either alone), completely blocking enamel cavitation, an outcome not observed before. Moreover, the anticaries e cacy was achieved at four times lower dosage of SnF 2 .
Notably, uoride, iron, and tin were detected in the outer layers of the enamel, indicating co-delivery to form a caries-protective lm in situ. Altogether, we developed a combination therapy with unexpected synergistic mechanisms that target the biological (bio lm) and physicochemical (enamel demineralization) traits simultaneously while providing a facile SnF 2 stabilization and lower dosage strategy against a widespread and costly oral disease, as summarized in Fig. 1.

Results
Antibio lm activity of Fer in combination with SnF 2 in vitro Fluoride is widely used as a gold standard anticaries agent, but it does not provide full protection, especially in severe cases where pathogenic bio lms rapidly accumulate. Despite its limited antibio lm activity, sodium uoride (NaF) can affect bacterial glycolysis and acid tolerance 20,21,22 , whereas stannous uoride (SnF 2 ) provides stronger antibacterial activity imparted by Sn 2+ ions 23,24 . First, we tested the antibio lm activity of both NaF and SnF 2 (1000 ppm of F, the typical concentration in over-thecounter formulations) and found that SnF 2 can signi cantly inhibit Streptococcus mutans (S. mutans, a cariogenic pathogen) viability and reduce the biomass more effectively than NaF (Fig. S1, A and B).
Afterward, we combined NaF or SnF 2 with Fer (1 mg of Fe/ml, an effective antibio lm concentration 17 ) in the presence of 1% H 2 O 2 (v/v). Remarkably, the combination of Fer with SnF 2 has substantially greater antibio lm activity than the combination of Fer and NaF (Fig. S1, C and D), resulting in no detectable viable bacteria and near complete biomass abrogation. In view of this result, we hypothesized that SnF 2 might be interacting with Fer for enhanced bioactivity.
We then investigated antibio lm activity using two dose-response studies using varying concentrations of Fer and SnF 2 . Given the potency of combination and high uoride (1000 ppm of F) concentration, we used the lowest dosage of SnF 2 (250 ppm of F) known to provide therapeutic effect as upper uoride dose limit. First, Fer dose (1 mg of Fe/ml) was xed and mixed with various concentrations of SnF 2 (0-250 ppm of F), and the number of viable cells and biomass were determined. As expected, Fer displayed a strong antibacterial effect against S. mutans bio lm (> 3-log reduction of viable cells; Fig. 2A), while also reducing biomass (Fig. 2B). When Fer was mixed with increasing concentrations of SnF 2 , both the antibacterial activity and the inhibitory effect on the biomass enhanced in a dose-dependent manner, indicating that SnF 2 can help improve the antibio lm e cacy of Fer. Next, the Fer concentration was varied (0-1 mg of Fe/ml) at constant SnF 2 dose (250 ppm of F). When combined with Fer, the antibacterial effect of SnF 2 increased in a dose-dependent manner, resulting in > 5-log reduction of viable cells compared to control when the concentration of Fer reached 1 mg of Fe/ml. Notably, the combination of Fer and SnF 2 was at least 2500-fold more effective in killing S. mutans cells than SnF 2 alone (Fig. 2C), suggesting a synergistic effect. We also found that SnF 2 (250 ppm of F) substantially reduces biomass ( Fig. 2D), although inclusion of increasing amounts of Fer did not enhance the bioactivity. The reduction of dry bio lm mass in response to SnF 2 treatment is likely due to the inhibition of secreted glucosyltransferases that are integral to the production of exopolysaccharides (EPS) by S. mutans, as reported by others 25 .
To further assess the antibio lm activity of the combination of Fer (1 mg of Fe/ml) and SnF 2 (250 ppm of F), confocal imaging was performed using uorescent labeling of the bacterial cells and α-glucan EPS. As depicted in Fig. 2E, the control bio lm contained bacterial clusters (in green) spatially arranged with abundant EPS (in red) matrix forming a densely packed structure. In a sharp contrast, the combination of Fer and SnF 2 impaired the accumulation of the bio lm where only small cell clusters with sparsely distributed EPS were detected. The orthogonal view images revealed that the spatial distribution of bacteria and EPS across the bio lm thickness was substantially compromised in the combination treated bio lm. These ndings were further con rmed by quantitative computational analyses, which showed that the combination of Fer and SnF 2 markedly reduced the biovolume of bacterial cells (Fig. 2F) and EPS ( Fig. 2G). SnF 2 stability in combination with Fer Given the enhanced e cacy of the combination of Fer and SnF 2 , we further investigated the physicochemical properties of this combination. The hydrodynamic diameter of Fer did not change signi cantly after adding SnF 2 (Table S1), indicating that SnF 2 is stable in the solution of Fer.
Additionally, we noticed that the zeta potential of Fer (Table S2) became less negative, serving as evidence that Fer interacts with SnF 2 , since coordination of Sn 2+ by the carboxylate groups is expected to lower the charge density of the carboxymethyl-dextran (CMD) corona. Representative transmission electron microscopy (TEM) images of Fer and Fer + SnF 2 after 1 h incubation in 0.1 M sodium acetate buffer (pH 4.5) are presented in Fig. S2. Consistent with dynamic light scattering (DLS) data, mixing Fer with SnF 2 did not seem to affect the size of Fer.
It is noteworthy that SnF 2 has limited stability in aqueous solutions owing to its high susceptibility to hydrolysis and oxidation 26, 27 requiring chemical additives (e.g., chelating agents) or anhydrous formulation 19 , which can reduce uoride bioavailability. We unexpectedly found that SnF 2 was stable in aqueous solutions containing Fer. To further investigate the stability of SnF 2 in the presence of Fer, SnF 2 (250 ppm of F) was mixed with increasing amounts of Fer in 0.1 M sodium acetate buffer at pH 4.5. We observed that the solution containing SnF 2 mixed with Fer was limpid after 24 h in sodium acetate buffer at pH 4.5 (Fig. 3A), demonstrating that Fer can enhance the stability of SnF 2 .
The enhanced stability of SnF 2 in the presence of Fer motivated us to investigate their chemical interactions. The core of Fer is coated with carboxymethyl-dextran (CMD) 28 . Thus, we explored whether SnF 2 can interact with CMD. SnF 2 alone or mixed with CMD was incubated in 0.1 M sodium acetate buffer (pH 5.5) for 24 h. We observed immediate formation of a precipitate when SnF 2 was dissolved in sodium acetate buffer, whereas the solution was limpid when mixed with CMD even after 24 h incubation ( Fig. 3B). In addition, the mixture of SnF 2 and CMD was characterized by 1 H nuclear magnetic resonance (NMR) spectroscopy (Fig. 3C). In the CMD spectrum, the anomeric proton (H1) in the C1 position was identi ed at 4.9 ppm, and protons (H2-H6) at the C2-C6 positions were detected at 3.2-4.0 ppm. The peak at 4.0-4.2 ppm (denoted as "a") is attributed to the protons of the carboxymethyl moieties as determined previously 29 . After CMD was mixed with SnF 2 , a clear shift in peak "a" was observed when compared to that of CMD alone. This suggests that Sn 2+ binds to the carboxymethyl moieties of CMD, which may account for the enhanced stability of SnF 2 with Fer. Note that similar 1 H NMR studies of SnF 2 and Fer are not possible due to the superparamagnetic nature of Fer interfering with 1 H NMR measurements.
In order to further investigate the effects of CMD on SnF 2 stability, we compared it with several control materials, i.e. dextran (Dex) (a similar polymer to CMD, but without carboxylic acid groups), as well as citric acid (CA), L-ascorbic acid (AA), and poly(acrylic acid) (PAA), which are all entities that all contain carboxylic acid groups. We found that Dex did not enhance the stability of SnF 2 , whereas each material that contains carboxylic acid groups did enhance stability (Fig. 3, D to F and Fig. S3). The Fer formulation also contains mannitol (Man) 28 , which is an antioxidant 30 . Since antioxidants can prevent the oxidation of SnF 2 31 , we added SnF 2 to various amounts of Man (1-10 mg/ml). Surprisingly, we did not observe any noticeable change in the stability of SnF 2 even with excess amounts of Man (10 mg/ml) (Fig. S4 To explore whether SnF 2 could in uence the catalytic activity of Fer, we used the 3,3′,5,5′tetramethylbenzidine (TMB) colorimetric assay for peroxidase-like activity following a previously published protocol 32 , with some modi cations. TMB is a chromogenic compound that yields a blue color upon oxidation with an absorption peak at 652 nm in the presence of reactive oxygen species (ROS), such as hydroxyl radical (•OH) 33 . As shown in Fig. 4, A and B, SnF 2 alone did not produce a noticeable amount of ROS. In contrast, the catalytic activity of Fer increased signi cantly after combining with SnF 2 as demonstrated by increased colorimetric reaction (Fig. 4, A and B), suggesting that SnF 2 enhanced the catalytic activity of Fer. Photographs in the inset of Fig. 4A exhibit the color change in each condition (SnF 2 , Fer, and Fer + SnF 2 from left to right, respectively).
Notably, we found that the enhancement of ROS production in the presence of SnF 2 is dependent on pH, concentration, and incubation time. The highest catalytic activity was observed at pH 4.5 (Fig. 4C). The greater ROS production at acidic pH conditions (characteristic of pathological conditions associated with dental caries) and the minimal ROS generation close to neutral (physiological) pH suggests a selectivity toward pathogenic bacteria. Surprisingly, very small amounts of SnF 2 are adequate for enhancing the catalytic activity of Fer (Fig. S5), whereby more ROS can be detected within 10 min incubation (Fig. S6A) gradually increasing to reach the highest level at 6 h ( Fig. S6B), which was maintained for prolonged period.
To further con rm the enhancement of the peroxidase-like activity of the Fer in the presence of SnF 2 , we employed a multi-pronged approach. First, we used OPD, a colorless substrate, which yields an oxidized product with a characteristic yellow color when reacting with ROS with an absorption peak at 450 nm 34 . As expected, the catalytic activity of Fer increased markedly after adding SnF 2 as compared to Fer alone and SnF 2 alone (Fig. 4D). We also measured ROS production via photoluminescence (PL) method using DCFH-DA as a ROS tracking indicator. DCFH-DA (a non uorescent molecule) yields a uorescent molecule DCF in the presence of ROS 35 . As depicted in Fig. 4E, the PL intensity increased to a greater extent after combining Fer with SnF 2 . Then, we measured the amount of hydroxyl radical (•OH) using coumarin as a photoluminescent probe molecule 36, 37 . As seen in Fig. 4F, Fer and SnF 2 in combination generated signi cantly more •OH than Fer alone, further demonstrating that SnF 2 enhanced the catalytic activity of Fer. In contrast, SnF 2 alone did not produce a noticeable amount of •OH (Fig. S7), consistent with its lack of catalytic activity.
Next, we investigated whether the augmented catalytic activity arises from different uoride or stannous salts. We replaced SnF 2 with NaF, a commonly used uoride salt in oral care formulations, or barium uoride (BaF 2 ), another uoride salt with a divalent cation of comparable size to Sn 2+ . We found that neither NaF nor BaF 2 increased the catalytic activity of Fer noticeably (Fig. 4, G and H), suggesting that F ─ may not play a crucial role in enhancing the ROS production performance of Fer. Conversely, we used SnCl 2 to evaluate whether Sn ions play a role in strengthening the ROS generation capability of Fer. We found SnCl 2 enhanced the catalytic activity of Fer (Fig. 4I), indicating that Sn ions may be playing a dominant role in increasing the catalytic performance of Fer. Taken together, these ndings support that SnF 2 can boost the catalytic ability of Fer, indicating that Fer and SnF 2 combination is an effective ROSgenerating therapy that can target bio lms under pathological (acidic) conditions.
We examined whether Fer released iron ions when combined with SnF 2 using inductively coupled plasma optical emission spectroscopy (ICP-OES). As depicted in Fig. S8A, the presence of SnF 2 slightly increased iron ions release from Fer at acidic pH (4.5). It is noteworthy that the amount of leached irons from Fer + SnF 2 formulation at circumneutral pH is negligible (Fig. S8B). Conversely, the iron leached from Fer + SnF 2 at acidic pH values could provide an added bene t. Iron ions have shown cariostatic effects as they can precipitate on the surface of enamel and promote the adsorption of phosphate and calcium ions, thereby reducing enamel demineralization 38, 39 .
Altogether, the increased stability of SnF 2 in aqueous solutions is mediated at least in part via interactions with CMD, which may be important for both uoride bioavailability and uoride delivery.
Unexpectedly, the presence of SnF 2 boosts the ROS generation capability of Fer at acidic pH, thus enhancing antibio lm e cacy under pathological condition. This synergistic Fer and SnF 2 combination provide a potent yet pH-dependent ROS-based therapy with enhanced antimicrobial uoride stability that could prevent the onset of dental caries in vivo.

Biocompatibility of Fer+SnF 2 in vitro
To examine whether this combination treatment is viable for use in vivo, the cytotoxicity of the combination of Fer and SnF 2 was assessed in human gingival keratinocytes (HGK) using MTS assay.
The cells were incubated with the combination of Fer (1 mg of Fe/ml) and SnF 2 (250 ppm of F) for 10 min, followed by 24 h incubation with fresh cell culture media. We found that the combined treatment of Fer and SnF 2 had no adverse effect on cell viability (Fig. S9).

Impact of Fer/SnF 2 on caries development and on enamel surface in vivo
Topical applications of Fer and SnF 2 in vivo were assessed using a rodent model that mimics the characteristics of severe human caries 40 , including sugar-rich diet and the development of surface zones 41 . Rat pups were infected with S. mutans (oral bacterial pathogen) and fed a sugar-rich diet (Fig. 5A). In this model, as depicted in Fig. 5B, tooth enamel progressively develops caries lesions (analogous to those observed in humans), proceeding from initial areas of demineralization to severe lesions characterized by enamel structure damage and cavitation. The test agents were topically applied twice daily with 1 min exposure time (Fig. 5A) to mimic the clinical use of a mouthwash. After the experimental period, the incidence and severity of caries lesions were evaluated. We also included a reduced concentration of the combination of Fer (0.25 mg of Fe/ml) and SnF 2 (62.5 ppm of F), since the lower amounts were capable of signi cantly killing the bacteria (p < 0.001) and reducing biomass (p < 0.001) compared to control in vitro (Fig. S10).
Quantitative caries scoring analyses revealed that the treatment of Fer in combination with SnF 2 was exceptionally effective in preventing caries development with higher e cacy than either alone (p < 0.001) (Fig. 5C). It nearly abrogated caries initiation and completely blocked further caries lesions development, thus preventing the onset of cavitation altogether (Fig. 5, D and E). The e cacy of the lower dosage of Fer and SnF 2 treatment was signi cantly greater than the control group (p < 0.001), and as effective as Fer (1 mg of Fe/ml) or SnF 2 (250 ppm of F) treatment alone. This demonstrates that the combination of Fer and SnF 2 has a synergistic effect for e cient bio lm treatment in vivo.
To determine the impact of treatment on the elemental composition of the enamel surface, lamellae oriented normal to the external enamel surface (EES) of rat mandibular rst molars (M1) were lifted out using a conventional focused ion beam (FIB) technique (Fig. S11). Line pro les normal to the EES were determined by scanning transmission electron microscopy (STEM) with energy dispersive spectroscopy (STEM-EDS) ( Fig. 5F) and aligned to the outer surface (see Methods). Comparing M1 from Fer + SnF 2 and control groups ( Fig. 5F(i)), we nd that the combined mole fractions of Fe, Sn, and F are substantially elevated, and the sum of the mole fractions Ca and P correspondingly reduced, in a thin lm at the surface. Preliminary analyses revealed that the thickness of this lm varies from ~ 50 to greater than 300 nm. Inspection of single element pro les from a 50 nm-thick lm ( Fig. 5F(ii to vi)) reveals that while the calcium mole fraction is reduced from ~ 25 at% to less than 5 at% in this layer, the mole fraction of oxygen remains at the same level as in the underlying enamel ( Fig. 5F(iii)). Sn reaches slightly more than 10 at% (Fig. 5F(vi)), while Fe is closer to 7 at% ( Fig. 5F(v)). While the pro les of the latter show broad maxima in the center of the layer, F levels appear highest at the outer surface at ~ 5 at% and decline to ~ 1 at% at the interface ( Fig. 5F(iv)). The presence of a Fe/Sn/F-rich layer was con rmed on a separately prepared sample from the same, treated rat molar using STEM with electron energy loss spectroscopy (STEM-EELS) (Fig. S12). Furthermore, the presence of Fe, Sn, and F at the surface of treated teeth, and the absence of these ions on vehicle-treated controls was con rmed using X-ray photoelectron spectroscopy (XPS, Fig. S13).
In high-resolution TEM (HRTEM) images, the Fe/Sn-rich layer was apparent as a slightly darker band between the enamel and the protective layers of FIB-deposited carbon and FIB-deposited Pt/C (Fig. S14). Lattice fringes were readily apparent in enamel (Fig. S14A), and Fast Fourier transform (FFT) images and radial integrals (Fig. S14, B and E) revealed sharp features consistent with the , , and sets of planes of crystalline hydroxylapatite. The Fe/Sn-rich layer did not display lattice fringes, and FFT images only showed diffuse scattering with a broad maximum at ~ 0.33 nm − 1 (Fig. S14, C and F), consistent with an amorphous oxide layer.
Taken together, there is strong evidence that there is a layer comprised of Fe and Sn with varying amounts of uoride present at the surface of the teeth of animals treated with SnF 2 + Fer. The presence of Ca and P in this layer may indicate co-precipitation during its formation. Gradients of Fe, Sn, and F at the interface between the lm and the underlying enamel suggest that these ions diffuse into enamel, but this remains to be con rmed.
Effect of Fer/SnF 2 on host microbiota and oral tissues in vivo The effects of Fer and SnF 2 on oral microbiota and surrounding soft tissues were also evaluated to assess the impact on oral microbiome diversity and oral tissue toxicity. All treatment groups showed no signi cant differences in alpha diversity among each group (Fig. 6, A  . Notably, microbiome data revealed a higher abundance of acidogenic bacterial genera such as Streptococcus and Lactobacillus in the control group, whereas they decreased in all the treatments (Fig. 6, C and D). Veillonella is known to consume acids produced by other acidogenic oral bacteria to grow and survive. Veillonella is especially reduced in the combined treatment groups, i.e. 1/4Fer + 1/4SnF 2 and Fer + SnF 2 (Fig. 6, C and D), indicating reduced acidogenic environment. In contrast, commensal genera related to oral health such as Haemophilus and Rothia were increased in treatment groups. Altogether, the microbiome data indicates that bacterial diversity is not affected as a community (Fig. 6D), but speci c bacteria associated with pathogenic environment are reduced by the combination treatment.
Histopathological analysis of gingival tissues revealed no indication of an acute in ammatory response, cytotoxicity, necrosis, or any changes in vascularization or proliferation, suggesting biocompatibility of Fer and SnF 2 treatment (Fig. 6E), consistent with in vitro data. Collectively, the data show that the combination was substantially more potent than either alone, whereas a lower concentration of agents in combination was as effective as each alone at full strength, indicating a synergistic effect between Fer and SnF 2 . In addition, the treatments did not disrupt the ecological balance of the oral microbiota or cause deleterious effects on the surrounding host tissues, indicating high precision for targeting cariogenic plaque-bio lms and preventing disease progression in vivo.

Discussion
In summary, we unexpectedly found a remarkable synergy between ferumoxytol (Fer) nanozymes and stannous uoride (SnF 2 ) in potentiating antibio lm and anticaries e cacy, which is particularly relevant given that current treatments are insu cient for controlling bio lm and preventing demineralization simultaneously in high-risk populations prone to disease. The combination treatment is far more effective than either alone and completely halts the progression of caries lesions and cavitation in a rodent model, without adverse effects on the surrounding host tissues or on the oral microbiota diversity in vivo.
Notably, we observed initial enamel lesions and eventually cavity formation when treated with SnF 2 or Fer alone, indicating that su cient acid is still generated to attack enamel. In sharp contrast, early lesions were seldom seen when treated with SnF 2 in combination with Fer. Furthermore, comparable therapeutic effects were achieved even at 4 times lower uoride concentration (62.5 ppm of F) when mixed with Fer, demonstrating the possibility of a therapy that uses very low doses (typical amounts in oral formulations ranges from 1000 to 1500 ppm of F). Such therapeutic synergy has not been observed previously in this animal model that mimics severe disease. Further analyses revealed that the improved effects achieved with the combination system can be attributed to three Our data indicate that Sn 2+ rather than F ─ is responsible for increasing the peroxidase-like activity of Fer.
It is possible that Sn 2+ in close proximity to the nanozyme core could e ciently accelerate the Fe 2+ /Fe 3+ redox cycles, while Sn-bound in the vicinity could serve as electron donors, resulting in electron transfer between Sn and Fe, thereby increasing ROS production. However, further studies are required to understand the exact mechanisms for catalytic enhancement in this system. Notably, the enhancement of catalytic activity was more pronounced at acidic pH value (4.5), typically found in cariogenic bio lms, whereas minimal ROS was generated close to neutral (physiological) pH, providing high selectivity and antibio lm activity. Taken together, it provides targeted activity under pathological conditions and operates at acidic pH values at which the anticaries action of SnF 2 is most effective 43 .
In addition to high antibio lm speci city and e cacy, the formation of an outer layer lm containing Fe, Sn, and F can provide a 'protective shield' against enamel acid demineralization. Fluoride acts by inhibiting mineral loss at the crystal surface and enhancing the rebuilding or remineralization of calcium and phosphate in a form more resistant to subsequent acid attacks 44 . The presence of a coating of metal-rich surface precipitate or a metal-rich surface layer can make enamel more acid-resistant 45,46 . We found a lm at the tooth surface that contains both Fe/Sn and F, and also variable amounts of calcium and phosphate. Calcium and phosphates contribute a protective role in preventing enamel demineralization by modulating physicochemical equilibrium and forming CaF 2 with uoride that reduces acid solubility while promoting remineralization 47 . To the best of our knowledge, the formation of Fe/Sn/F polyion lm has not been described previously and potentially a novel mechanism for caries prevention.
Despite promising results, there are some limitations, but also opportunities for further research. Although our preliminary study suggests that carboxylates play an integral role in enhancing the stability of SnF 2 , additional analyses are needed to understand the physicochemical interactions between SnF 2 and Fer as well as the long-term stability of the complexes and the oxidation state of Sn in the complexes. Additional studies are required to elucidate the exact mechanisms by which ROS generation is enhanced by SnF 2 .
Further analyses on how the metal ion-uoride lm is formed may reveal additional insights on the enamel remineralization process. Additionally, full toxicity studies are needed to determine the long-term effects of daily use of Fer and SnF 2, whereas optimization of the concentrations of Fer, SnF 2 , and H 2 O 2 may be required for clinical translation and product development. Nevertheless, our data reveal that Fer and SnF 2 potentiate the therapeutic activity through unexpected synergistic mechanisms that target both the biological (bio lm) and physicochemical (enamel demineralization) traits of dental caries simultaneously.
This simple yet effective combination therapy with uoride co-delivery could advance current anticaries treatment while leading to the development of ROS-based modalities for other bio lm-related diseases.
The search for new modalities encompasses novel compounds, where further development involves a lengthy and costly process and regulatory approval. The ndings that an off-the-shelf iron oxide nanoparticle formulation has a potent topical effect at a fraction (< 0.2%) of the approved systemic dosage together with low dose of SnF 2 that operates through complementary mechanisms of action can facilitate its path to clinical translation. This approach could be targeted for high-risk individuals prone to cariogenic bio lm accumulation without increasing the risk of uoride overexposure. It is noteworthy that patients with severe childhood tooth decay is often linked with iron de ciency anemia 38, 48, 49, 50 . The possibility that two major global health problems, i.e., tooth decay and anemia 50, 51 , could be treated by using Fer and SnF 2 opens a feasible opportunity to include the combination therapy in clinical trials for caries prevention tailored to high-risk patients with iron-de ciency anemia.

Methods
In vitro bio lm model and quantitative analysis Bio lms were formed using the saliva-coated hydroxyapatite disc (sHA) model as described elsewhere 17,32,52 . S. mutans UA159, a proven virulent and well-characterized cariogenic pathogen, was grown in ultra- ROS measurement using 3,3',5,5'-tetramethylbenzidine (TMB) assay The catalytic activity of Fer + SnF 2 was investigated by a colorimetric assay using TMB (Sigma-Aldrich) as a probe, which generates a blue color after reacting with ROS 33  ROS study using 2 ,7 -dichloro uorescin diacetate (DCFH-DA) probe In order to further support the enhancement of the catalytic activity of Fer in the presence of SnF 2 , we used photoluminescence (PL) method using DCFH-DA (Sigma-Aldrich) as a ROS probing agent 35

Iron release study
The release of soluble iron from Fer, in the presence and absence of SnF 2 , was investigated using inductively coupled plasma optical emission spectroscopy (ICP-OES, Spectro Genesis). Brie y, 10 ml of Fer (0.5 mg of Fe/ml) was incubated with or without SnF 2 (0.5 mg/ml) for 1 h in 0.1 M sodium acetate buffer (pH 4.5, 5.5, or 6.5) at room temperature. Afterward, free iron ions and intact nanoparticles were separated by centrifugation using ultra ltration tubes (3 kDa, MWCO). The pellet was then resuspended in the same volume using 0.1 M sodium acetate buffer. Subsequently, the ltrate and resuspend pellet were digested in nitric acid and nally diluted with DI water before analysis by ICP-OES.
Toxicity study of the combined treatment of Fer and SnF 2 in human gingival keratinocytes (HGK) The in vitro biocompatibility of the combination of Fer and SnF 2 was investigated in HGK cells using an After that, the media was removed, the cells were washed twice with sterile phosphate buffered saline (PBS) and 100 µl of fresh complete cell culture media was added to each well. After 24 h incubation, the cell culture media was removed, and 20 µl of MTS reagent and 100 µl of media were added to each well. After 3 h additional incubation under standard cell culture conditions, the absorbance was recorded at 490 nm using a microplate reader. The cell viability was calculated using the following formula: In vivo e cacy of Fer in combination with SnF 2 In vivo e cacy was assessed using a well-established rodent model of dental caries, as reported previously 40,53 . In brief, 15 days-old speci c pathogen free Sprague-Dawley rat pups were purchased with their dams from Harlan Laboratories (Madison, WI, USA). Upon arrival, animals were screened for S. mutans by plating oral swabs on mitis salivarius agar plus bacitracin (MSB). Then, the animals were orally infected with S. mutans UA159, and their infections were con rmed at 21 days via oral swabbing.
To simulate a clinical scenario, a topical treatment regimen was used that consisted of a short exposure proceeded for 5 weeks, and their physical appearance was recorded daily. At the end of the experimental period, all animals were sacri ced, and their jaws were surgically removed and aseptically dissected, followed by sonication to recover total oral microbiota as reported previously 54 . All of the jaws were de eshed, and the teeth were prepared for caries scoring based on Larson's modi cation of Keyes' system 40 . Determination of the caries score of the jaws was performed by a calibrated examiner who was blinded for the study by using codi ed samples. Enamel surfaces were analyzed as described below. Moreover, the gingival tissues were collected for hematoxylin and eosin (H&E) staining for histopathological analysis by an oral pathologist at Penn Oral Pathology. This research was reviewed and approved by the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC #805529).

Preparation of enamel samples for scanning transmission electron microscopy (STEM)
We identi ed the most promising location for focused ion beam (FIB) lift-out as the middle cusp of the buccal side by assessing curvature and roughness using synchrotron micro-computed tomography identi ed manually from line pro les. Pro les were aligned on the outer surface position, and the distance axis was set to zero at the interface between the Fe and Sn rich layer and enamel. Data were plotted as the mean value at the given distance (solid circles), and in smoothed form (lines), as the local 3-point mean (moving average with span 3, using the movmean() function).
EEL spectra were acquired with a GIF continuum system (Gatan) using a K3 IS direct electron detector (Gatan) in counting mode at 300 kV. The high quantum e ciency of this detector (DQE up to 90%) allowed the simultaneous acquisition of the relevant inner shell ionization (core loss) edges and zero loss region at high energy resolution, except for the phosphorous K and L edges, which were outside the selected energy range. The convergence semi-angle of the probe was 19 mrad, and the probe current was 27 pA, as determined using a Faraday cup. The collection semi-angle of 36 mrad was de ned by the EELS entrance aperture (5 mm). The three-dimensional spectrum image dataset was collected using an energy dispersion of 0.35 eV/channel and the probe dwell time was 4 ms/pixel with a pixel size of 6 nm, with sub-pixel scanning enabled (32 × 32) to yield a ~ 3.8 Å pixel. Simultaneously, ADF images were acquired using a collection semi-angle of 51-115mrad. In post-processing, the zero-loss peak was aligned in every pixel of the spectrum image using GMS software (Gatan, Inc). Elemental Quanti cation Analysis was performed in the same software, using a Hartree-Slater cross-section model and including plural scattering corrections.
High-resolution TEM (HRTEM) imaging HRTEM imaging of enamel specimens was performed using an JEOL GrandARM 300F at an accelerating voltage of 300 kV. Images (edge length: 4096 pixels, scale factor 0.0328 nm/pixel) were processed using Matlab r2022b (Mathworks, Natick, MA). Two-dimensional Fourier transforms of regions of interest (edge length: 1024 pixels) were determined using fft2() and rearranged using fftshift() to move the zero frequency components to the center of the image. Fourier transform images were unwrapped in the azimuthal direction by interpolation using griddedInterpolant() with a query grid in polar coordinates (radial pitch: 0.0298 nm − 1 /pixel; azimuthal pitch: 1˚/pixel) and integrated in the azimuthal direction.

X-ray photoelectron spectroscopy (XPS)
Two mandibular (M1) rat molars, one from Fer + SnF 2 treated group and one from control group, were dissected and attached using copper tape (Electron Microscopy Sciences). XPS analysis was conducted using a Thermo Scienti c Nexsa G2 using an Al-Ka X-ray source, with the following parameters: pressure of 2·10 − 9 torr (2.5·10 − 7 Pa), an X-ray gun power of 150 W, a spot diameter of 100 µm, and a takeoff angle of 0º. XPS survey spectra were acquired under a pass energy of 100 eV, using a step size of 1 eV. Highresolution spectra for F, Fe, Ca, P, O, Sn, Na, and Mg were acquired under a pass energy of 50 eV, using a step size of 0.1 eV, and averaging over 10 scans. For depth pro ling, the surface was excavated using an argon ion beam (4 keV, diameter 500 µm, 'high current' mode, 30-300s increment) between successive spectra. All data were processed using Avantage (Thermo Scienti c), and spectra were referenced to adventitious carbon at 284.8 eV.
16S rRNA sequencing Cells were pelleted from dental plaque by centrifuging at maximum speed for 5 min. DNA was extracted from the pellets using the Qiagen DNeasy PowerSoil htp kit according to the manufacturer's instructions within a sterile class II laminar ow hood. Mock washes and mock extractions were included to control for microbial DNA contamination arising through the sonication and extraction processes, respectively.
Polymerase chain reaction (PCR) ampli cation of V1-V2 region of 16S rRNA gene was performed using Golay-barcoded universal primers 27F and 338R. Four replicate PCR reactions were performed for each sample using Q5 Hot Start High Fidelity DNA Polymerase (New England BioLabs). Each PCR reaction contained: 4.3 µl microbial DNA-free water, 5 µl 5X buffer, 0.5 µl dNTPs (10 mM), 0.17 µl Q5 Hot Start Polymerase, 6.25 µl each primer (2µM), and 2.5 µl DNA. PCR reactions with no added template or synthetic DNAs were performed as negative and positive controls, respectively 55 . PCR ampli cation was done on a Mastercycler Nexus Gradient (Eppendorf) using the following conditions: DNA denaturation at 98 ºC for 1 min, then 20 cycles of denaturation at 98 ºC for 10 sec, annealing 56 ºC for 20 sec and extension 72 ºC for 20 sec, last extension was at 72 ºC for 8 min. PCR replicates were pooled and then puri ed using a 1:1 ratio of Agencourt AMPure XP beads (Beckman Coulter, Indianapolis, IN), following the manufacturer's protocol. The nal library was prepared by pooling 10 µg of ampli ed DNA per sample. Those that did not arrive at the DNA concentration threshold (e.g., negative control samples) were incorporated into the nal pool by adding 12 µl. The library was sequenced to obtain 2x250 bp paired-end reads using the MiSeq Illumina 56 .
To analyze 16S RNA gene sequences, we used QIIME2 v19.4 57 . We obtained taxonomic assignments based on GreenGenes 16S rRNA database v.13_8 58 and ASV analysis of shared and unique bacterial taxa through DADA2 59 . PCoA was performed using library ape for R programming language 60 . To test the differences between communities, we used library vegan and UniFrac distances (https://CRAN.Rproject.org/package=vegan). R environment (version 4.0.3) was used for statistical analysis. Nonparametrical test Wilcoxon Rank Sum Test was performed for the pairwise comparison between treatment groups for richness and Shannon diversity analysis. PERMANOVA analysis was performed for weighted UniFrac principal coordinate analysis to evaluate the differences between treatment groups.

Statistical analysis
The data presented as the mean ± standard deviation were performed at least three times independently unless otherwise stated. One-way analysis of variance (ANOVA) followed by the Tukey test was used to determine the statistical signi cance between the control and the experimental groups unless otherwise stated. p values < 0.05 were considered statistically signi cant. is bound by carboxylate groups in the carboxymethyl-dextran coating of Fer. (E) Using laboratory and in vivo models, we nd synergistic activities to enhance bioactivity against bio lms and caries-protective effects (without increasing uoride exposure), while co-delivering uoride, iron, and tin on the outer enamel surface without deleterious effects on oral tissues and the microbiota.   The data are presented as mean ± standard deviation. ***p < 0.001; ns, nonsigni cant; one-way ANOVA followed by Tukey test. (vi) vs. distance in the direction normal to the EES for M1 rat molars from rates treated with Fer+SnF 2 and untreated controls. Pro les were manually aligned on the outer surface, and that the distance axis is referenced to the approximate position of the interface between the Fe/Sn/F-rich layer and the underlying enamel of the treated sample.