Synergistic antibacterial effect of photocatalytic coatings activated by visible light and photodynamic therapy for rapid peri-implantitis treatment.

Bruna Egumi Nagay: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Conceptualization. Raphael Cavalcante Costa: Writing – review & editing, Methodology, Investigation. Samuel Santana Malheiros: Writing – review & editing, Methodology, Investigation. Jairo Matozinho Cordeiro: Writing – review & editing, Methodology, Investigation. Caroline Dini: Writing – review & editing, Methodology, Investigation. Ariane Bezerra Santos: Writing – review & editing, Methodology, Investigation. Cristiane Cantiga: Writing – review & editing, Methodology, Investigation. Barbara Ribeiro Rios: Writing – review & editing, Methodology, Investigation. Xiang Li: Investigation, Methodology, Writing – review & editing. Zhang Ruixin: Investigation, Methodology, Writing – review & editing. Xi Rao: Investigation, Methodology, Writing – review & editing. Camila Schmidt Stolf: Writing – review & editing, Methodology, Investigation. Renato Corrêa Viana Casarin: Writing – review & editing, Methodology, Investigation. Nilson Cristino da Cruz: Writing – review & editing, Methodology, Investigation. Fabiane Hiratsuka Veiga de Souza: Writing – review & editing, Methodology, Investigation. Luciano Tavares Angelo Cintra: Writing – review & editing, Methodology, Investigation. Sathishkumar Gnanasekar: Writing – review & editing, Methodology, Investigation. Selvakumar Murugesan: Writing – review & editing, Methodology, Investigation. Liqun Xu: Writing – review & editing, Methodology, Investigation. Leonardo Perez Faverani: Writing – review & editing, Methodology, Investigation. Jeroen J.J.P. van den Beucken: Writing – review & editing, Methodology, Investigation. Valentim A.R. Barão: Writing – review & editing, Supervision, Funding acquisition, Conceptualization.

All human-related procedures were conducted in accordance with the Declaration of Helsinki and were approved by the Ethics Committee of the Piracicaba Dental School, University of Campinas (UNICAMP), under protocol 19100719.0.0000.5418. Documentation of this approval is available through Plataforma Brasil ( https://plataformabrasil.saude.gov.br/login.jsf ). Written informed consent was obtained from all participants prior to enrollment. All animal experiments were performed in compliance with institutional and national guidelines for the care and use of laboratory animals. The study protocol was reviewed and approved by the Ethics Committee of the Araçatuba Dental School, São Paulo State University (UNESP), under registration number 142-2023.

The PEO process was performed by immersing cpTi discs in an electrolytic solution containing nitric acid and ammonium acetate as conductive agents to promote the formation of a TiO 2 coating. To obtain Bi-doping in TiO 2 coatings, bismuth nitrate precursor was added (Bi-TiO 2 ; for details, see Experimental Section; Scheme 1 A). Commercially pure titanium (cpTi) discs with a polished surface were selected as controls to provide a standardized substrate, ensuring that the effects observed could be attributed solely to the coating. Scanning electron microscopy combined with energy-dispersive X-ray spectroscopy (SEM-EDS) and confocal laser scanning microscopy (CLSM) images of the surfaces confirmed the successful formation of the coatings on Ti substrates ( Fig. 1 A–C). The physicochemical surface features were similar to those in our previous study [ 16 ]. After PEO treatment, we observed that the Ti surface was modified entirely due to the growth of a sponge-like coating. Notably, the SEM images showed that the surface morphology was slightly affected by the presence of Bi. The TiO 2 coating exhibited craters surrounded by irregular, randomly distributed features. With the incorporation of Bi — confirmed by energy-dispersive spectroscopy ( Fig. 1 A, Fig. S1–S2 ), which revealed an atomic concentration of Bi of 3.65 ± 1.54 at% in the coating ( Table S1 ) – these features persisted but appeared less frequently, along with the emergence of round-shaped structures of varying diameters, contributing to a slight decrease in surface area (Bi-TiO 2  = 2.24 cm 2 ; TiO 2  = 2.53 cm 2 ; CLSM; Fig. 1 A–C). The incorporation of bismuth ions (Bi 3+ ) and changes in surface morphology may be attributed to the use of bismuth nitrate as a precursor, which is known to be a powerful oxidizing agent [ 21 ]. When bismuth nitrate is introduced into water, it dissociates and generates Bi 3+ . Then, when the breakdown voltage threshold during the PEO process is surpassed, the coating becomes porous, and Bi incorporation is intensified due to plasma discharges triggered by fluctuations in molten titanium oxide on the anode surface. As the voltage continues to rise, the incorporation of stable Bi 3+ with oxidation states such as Bi(OH) 3 and BiO becomes more prevalent, enhancing the coating's compactness. This increased compactness enables the coating to withstand strong electric fields, thereby reducing detrimental effects and generating less powerful plasma discharges. Consequently, the resulting coating acquires a hybrid structure, characterized by a denser inner region and a more porous outer layer. This outer architecture exhibited smaller pores and reduced overall porosity, along with the presence of Bi nucleation sites on the external surface [ 22 , 23 ]. Concurrently, as the microdischarge channels experience high temperatures followed by the melting-cooling process, the growth of an inorganic layer consisting of crystalline oxides of the (here titanium) substrates occurs [ 24 ]. We further substantiated the coating formation by XRD spectra ( Fig. 1 D), which displayed distinct diffraction peaks characteristic of both the anatase and rutile phases, confirming the crystalline formation of TiO 2 [ 16 ]. Such a unique mixed crystalline coating configuration exhibits relatively enhanced carrier transfer in photocatalytic activity and features intensified three-dimensional lattices of metallic atoms, which are favorable for increased hardness and wear resistance behavior. Fig. 1 Physicochemical characterization of the coatings. (A–C) Representative top-view SEM micrographs and CLSM 3D topography images, with corresponding EDS elemental mapping, for (A) cpTi, (B) TiO 2 , and (C) Bi–TiO 2 surfaces. (D) XRD patterns (A = anatase; R = rutile; Ti = titanium). (E–G′) XPS high-resolution spectra: (E) O1s (cpTi), (E′) O1s (TiO 2 ), (E″) O1s (Bi–TiO 2 ); (F) Ti2p, Bi4d (cpTi), (F′) Ti2p, Bi4d (TiO 2 ), (F″) Ti2p, Bi4d (Bi–TiO 2 ); (G) Bi 4f (Bi–TiO 2 ), and (G′) N1s (Bi–TiO 2 ). (H) Water contact angle. (I) Cumulative Bi ion release in PBS (ppm). (J) Surface roughness (profilometry). Quantitative data are presented as mean ± SD (n = 5). Statistical comparisons of panels H and J were performed using one-way ANOVA followed by Tukey's post hoc test (∗∗∗∗P < 0.0001). cpTi: commercially pure titanium; TiO 2 : titanium dioxide; Bi–TiO 2 : bismuth-modified TiO 2 coating. Fig. 1 Physicochemical characterization of the coatings. (A–C) Representative top-view SEM micrographs and CLSM 3D topography images, with corresponding EDS elemental mapping, for (A) cpTi, (B) TiO 2 , and (C) Bi–TiO 2 surfaces. (D) XRD patterns (A = anatase; R = rutile; Ti = titanium). (E–G′) XPS high-resolution spectra: (E) O1s (cpTi), (E′) O1s (TiO 2 ), (E″) O1s (Bi–TiO 2 ); (F) Ti2p, Bi4d (cpTi), (F′) Ti2p, Bi4d (TiO 2 ), (F″) Ti2p, Bi4d (Bi–TiO 2 ); (G) Bi 4f (Bi–TiO 2 ), and (G′) N1s (Bi–TiO 2 ). (H) Water contact angle. (I) Cumulative Bi ion release in PBS (ppm). (J) Surface roughness (profilometry). Quantitative data are presented as mean ± SD (n = 5). Statistical comparisons of panels H and J were performed using one-way ANOVA followed by Tukey's post hoc test (∗∗∗∗P < 0.0001). cpTi: commercially pure titanium; TiO 2 : titanium dioxide; Bi–TiO 2 : bismuth-modified TiO 2 coating. Besides higher content of oxygen observed in EDS ( Table S1 ), X-ray photoelectron spectroscopy (XPS) analysis further confirmed the presence of a TiO 2 layer and the presence of Bi in the form of Bi 3+ in Bi-TiO 2 coating ( Fig. 1 E–G), as indicated by the overlap between the Ti 2p 1/2 and Bi 4d 3/2 peaks (at ∼464.5 eV). In the high-resolution XPS spectrum, peaks at the characteristic spin-orbit doublet pattern of Ti 2p 3/2 and Ti 2p 1/2 exhibited energies of 457.3 eV and 463.0 eV, indicating the presence of Ti(IV) oxidation state, corresponding to TiO 2 [ 22 ], whereas peaks at 441.4 and 465.4 eV were attributed to the Bi 4d 5/2 and Bi 4d 3/2 orbitals, respectively. Additionally, the distinct doublet components at Bi 3d 5/2 , 4f 5/2 , and 4f 7/2 with peaks at 464.3, 159.4, and 164.6 eV in the high-resolution Bi 4d and Bi 4f spectra are indicative of trivalent Bi cations [ 22 , 25 ]. Semi-quantitative XPS analysis further revealed a surface Bi atomic concentration of 9.37 ± 0.67 at%, confirming the enrichment of Bi species at the outermost layer of the coating ( Table S1 ). Notably, the analysis of the O1s spectra revealed three distinct peaks, with energies at approximately 529.8 eV (representing lattice oxygen), 530.7 eV (indicating oxygen defects), and 531.9 eV (suggesting the presence of hydroxyl groups formed due to surface-adsorbed water contact). The appearance of Bi-OH peaks can be attributed to two primary factors: i) dissociation of bismuth nitrate in water, generating Bi 3+ ions, which then react with hydroxide ions (OH − ) from the basic electrolyte solution, or ii) interaction with OH − originating from a secondary source, linked to the reduction of oxygen (O 2 ), nitrate ions (NO 3 − ), and water (H 2 O) [ 23 ]. Furthermore, the rapid increase in temperature and pressure within the discharge channels during the PEO process has likely resulted in the partial oxidation of Bi(OH) 3 , leading to the formation of BiO, which explains the observed Bi-O peak in the Bi 4f spectra. In addition, the analysis of the N1s spectra ( Fig. 1 G’) revealed three components centered at approximately 398.1, 403.3, and 412.6 eV. The peak at ∼398.1 eV can be attributed to molecular nitrogen from the dissolution of electrolyte precursors (nitric acid and bismuth nitrate), while the component at ∼403.3 eV is consistent with oxidized Ti–N–O configurations, suggesting minimal incorporation of nitrogen during the high-energy micro-arc discharges of the PEO process. The high-binding-energy peak at ∼412.6 eV, with lower intensity, is more plausibly assigned to residual nitrate species derived from the electrolyte precursors [ 26 , 27 ]. Collectively, these features indicate that nitrogen is mainly associated with reaction byproducts and electrolyte-derived species rather than representing a functional N doping of the Bi–TiO 2 coating. PEO-related microdischarges characterized by high-energy particle bombardment are likely to have introduced oxygen vacancies (OVs) into the TiO 2 lattice [ 28 ], as indicated here by the binding energy signal at 531.2 eV in the O1s spectra. This process probably led to higher concentrations of Ti 3+ due to the reduction of Ti 4+ by introducing OVs. These localized OVs are known to transfer electrons to neighboring Ti 4+ atoms, converting them into Ti 3+ . Consequently, this phenomenon disrupts the bond between Ti and lattice oxygen, forming new hydroxyl groups that are individually bonded to Ti atoms, which renders the surface more hydrophilic. This mechanism thus explains the superhydrophilic nature of TiO 2 and increased hydrophilicity of Bi-TiO 2 compared to the cpTi surface ( Fig. 1 H), which is known to favor interactions with cells in terms of attachment and spreading [ 29 ]. In particular, previous studies have provided evidence that hydrophilic surfaces and/or those featuring Bi 3+ on their surface facilitate cell adhesion and tissue repair [ 30 , 31 ]. Given that Bi 3+ in the Bi–TiO 2 coating can be both nucleated at the surface and gradually released into the surrounding medium, its release profile was also determined. As depicted in Fig. 1 I, Bi ions exhibited continuous release throughout the testing period. The initial release rate (up to 4 h) was relatively high, followed by a slower release rate (up to 7 days), indicating the chemical stability and sustained release of Bi ions from this coating. It is important to note that bismuth is generally regarded as safe, with a toxic intake level estimated at approximately 15 g for a 70 kg human [ 32 , 33 ]. Notably, the concentration of Bi ions released from the coating in this study remained well below this threshold. Bi is widely used in pharmaceuticals as a "green" heavy metal due to its anti-cancer, anti-inflammatory, and antibacterial properties against Gram-positive and Gram-negative bacterial strains. These properties could also contribute to the control of peri-implant infections. In addition to hydrophilic and chemical stability characteristics, it is widely recognized that a certain level of surface roughness is favorable for implants [ 34 ], as it significantly impacts cell adhesion, proliferation, and differentiation. Accordingly, we evaluated the roughness of the surface using profilometry. As shown in Fig. 1 J, surface modification with PEO treatment significantly increases surface roughness, which is consistent with the higher surface area observed in CLSM images ( Fig. 1 A–C). Microscale surface roughness can enhance cell adhesion and osseointegration by stabilizing fibrin clots through physical interlocking, which anchors clotting elements and guides the growth of bone-forming cells at the implant–bone interface [ 35 ]. This direct attachment can enhance the bond between the implant and bone, favoring implant stability. It is well established that implant survival and success are also linked to physical stability, hardness, and long-term durability. Implants are intermittently subjected to load bearing during mastication [ 36 ]. At the dental implant surface level, this can lead to surface wear and degradation, releasing particles that serve as risk indicators for the development of peri-implant infections [ 37 ], and, hence, implant failure. To investigate whether the Bi-TiO 2 coating contributes to a more protective surface against wear, we conducted tribological testing using a pin-on-disk tribological system [ 38 ]. The dynamic friction coefficient curves were tested at 0.01 m s −1 under a 5 N load, with simulated body fluid (SBF) lubrication. Fig. 2 A shows that the friction coefficient curves recorded during this test exhibited a visible difference between the cpTi and PEO coatings. While the cpTi controls displayed a constant friction coefficient curve at approximately 0.2, both TiO 2 and Bi-TiO 2 coatings showed a higher friction coefficient of about 0.3, with the curves following an accordion-like characteristic with fluctuations. The higher friction coefficient can be primarily attributed to the accumulation and confinement of numerous wear particles, often referred to as "third bodies," within the contact region. Also, the total mass loss after testing was markedly reduced for TiO 2 and Bi–TiO 2 compared with the control, indicating mitigated material removal under localized mechanical loading ( Fig. 2 B). This behavior can be explained by the microhardness results, as both PEO-derived coatings presented higher Vickers hardness than cpTi, supporting an enhanced resistance to wear initiation ( Fig. 2 C) promoted by a more crystalline structure of the surface [ 39 ]. Corroborating these quantitative findings, SEM micrographs and optical microscopy revealed that cpTi surfaces showed extensive wear scars and broader track areas, whereas TiO 2 and Bi–TiO 2 layers remained present after testing and displayed narrower, less severe wear tracks ( Fig. 2 D and E). Fig. 2 Mechanical and tribological behavior of the coatings. (A) Friction coefficient profile and (B) Total mass loss after the tribological test. (C) Vickers microhardness of all groups. (D) Wear-track characterization by optical photomicrographs and SEM images. (E) Schematic illustration of the proposed tribological protection features provided by the Bi–TiO 2 coating. Quantitative data are presented as mean ± SD (n = 5). Statistical comparisons for (B–C) were performed using one-way ANOVA followed by Tukey's post hoc test (∗P < 0.05, ∗∗P < 0.01, ∗∗∗∗P < 0.0001). cpTi: commercially pure titanium; TiO 2 : titanium dioxide; Bi–TiO 2 : bismuth-modified TiO 2 coating. Illustration created with BioRender (Barao, V.; 2025). Fig. 2 Mechanical and tribological behavior of the coatings. (A) Friction coefficient profile and (B) Total mass loss after the tribological test. (C) Vickers microhardness of all groups. (D) Wear-track characterization by optical photomicrographs and SEM images. (E) Schematic illustration of the proposed tribological protection features provided by the Bi–TiO 2 coating. Quantitative data are presented as mean ± SD (n = 5). Statistical comparisons for (B–C) were performed using one-way ANOVA followed by Tukey's post hoc test (∗P < 0.05, ∗∗P < 0.01, ∗∗∗∗P < 0.0001). cpTi: commercially pure titanium; TiO 2 : titanium dioxide; Bi–TiO 2 : bismuth-modified TiO 2 coating. Illustration created with BioRender (Barao, V.; 2025). In PDT, ROS are generated through visible-light activation of a photosensitizer. The localized production of ROS at bacterially infected sites plays a central role in microbial inactivation and represents the primary mechanism of PDT in peri-implant infections. However, PDT alone often shows variable efficacy, highlighting the need for strategies capable of enhancing ROS generation. In this context, we hypothesized that integrating a Bi–TiO 2 coating with PDT could potentiate oxidative effects, given its visible-light photocatalytic activity and spectral compatibility with methylene blue (MB). To explore this hypothesis, the optical response of TiO 2 and Bi–TiO 2 surfaces was evaluated using UV–visible diffuse reflectance spectroscopy (DRS). As shown in Fig. 3 A, TiO 2 exhibited an absorption edge at approximately 450 nm, whereas Bi–TiO 2 displayed a pronounced red shift extending to around 560 nm, indicating enhanced visible-light absorption under the safe LED irradiation used in this study (λ = 420–690 nm; main emission peak ∼560 nm) ( Fig. S3 ). Bandgap energies (Eg), estimated via the Kubelka–Munk function (αhν = A(hν – Eg) n ), were determined as 2.67 eV for TiO 2 and 2.14 eV for Bi–TiO 2 ( Fig. 3 B). This bandgap narrowing explains the red shift and supports effective visible-light activation of Bi–TiO 2 . The same behavior can be observed from the VB XPS spectra ( Fig. 3 C), in which the reduced energy separation between the VB edge and the Fermi level in Bi–TiO 2 indicated an upward VB shift of 0.86 eV, reflecting Bi-induced band structure modulation. Fig. 3 Optical, photoelectrochemical, and ROS characterization of TiO 2 and Bi–TiO 2 coatings. (A) UV–vis diffuse reflectance spectra (DRS). (B) Tauc plots derived from the Kubelka–Munk function used to estimate bandgap energies (Eg). (C) Valence band XPS spectra indicating the shift of the valence band (VB) edge relative to the Fermi level. (D,E) Mott–Schottky plots of TiO 2 and Bi–TiO 2 coatings at different frequencies. (F) Transient photocurrent (I-t) responses under repeated light on/off cycles. (G) Nyquist plots obtained by electrochemical impedance spectroscopy (EIS). (H) Photocatalytic activity evaluated by methylene blue (MB) degradation under visible-light irradiation. (H′) Fold-change in MB degradation relative to cpTi. (I) Total ROS concentration under dark and light conditions, as determined by electron paramagnetic resonance (EPR) spectroscopy. (J–L) EPR spectra for selective ROS detection: total ROS (J), hydroxyl radicals (•OH) (K), and superoxide radicals (1O 2 ) (L). (M) Band structure diagram illustrating conduction band (CB) and valence band (VB) positions relative to redox potentials. (N) Schematic representation of the synergistic ROS-generation mechanism combining Bi–TiO 2 photocatalysis and methylene blue-mediated photodynamic therapy (PDT) under visible-light irradiation. Quantitative data are presented as mean ± SD (n = 5). Statistical comparisons for (H′ and I) were performed using one-way ANOVA followed by Tukey's post hoc test in each light condition (∗P < 0.05, ∗∗P < 0.01, ∗∗∗∗P < 0.0001). cpTi: commercially pure titanium; TiO 2 : titanium dioxide; Bi-TiO 2 : titanium dioxide doped with bismuth. Illustration created in BioRender. Barao, V. (2025). Fig. 3 Optical, photoelectrochemical, and ROS characterization of TiO 2 and Bi–TiO 2 coatings. (A) UV–vis diffuse reflectance spectra (DRS). (B) Tauc plots derived from the Kubelka–Munk function used to estimate bandgap energies (Eg). (C) Valence band XPS spectra indicating the shift of the valence band (VB) edge relative to the Fermi level. (D,E) Mott–Schottky plots of TiO 2 and Bi–TiO 2 coatings at different frequencies. (F) Transient photocurrent (I-t) responses under repeated light on/off cycles. (G) Nyquist plots obtained by electrochemical impedance spectroscopy (EIS). (H) Photocatalytic activity evaluated by methylene blue (MB) degradation under visible-light irradiation. (H′) Fold-change in MB degradation relative to cpTi. (I) Total ROS concentration under dark and light conditions, as determined by electron paramagnetic resonance (EPR) spectroscopy. (J–L) EPR spectra for selective ROS detection: total ROS (J), hydroxyl radicals (•OH) (K), and superoxide radicals (1O 2 ) (L). (M) Band structure diagram illustrating conduction band (CB) and valence band (VB) positions relative to redox potentials. (N) Schematic representation of the synergistic ROS-generation mechanism combining Bi–TiO 2 photocatalysis and methylene blue-mediated photodynamic therapy (PDT) under visible-light irradiation. Quantitative data are presented as mean ± SD (n = 5). Statistical comparisons for (H′ and I) were performed using one-way ANOVA followed by Tukey's post hoc test in each light condition (∗P < 0.05, ∗∗P < 0.01, ∗∗∗∗P < 0.0001). cpTi: commercially pure titanium; TiO 2 : titanium dioxide; Bi-TiO 2 : titanium dioxide doped with bismuth. Illustration created in BioRender. Barao, V. (2025). Then, because photocatalytic ROS generation depends on band structure, Mott–Schottky (M − S) analyses were performed. Both coatings exhibited positive slopes ( Fig. 3 D and E), confirming n-type semiconductor behavior. The flat-band potentials were −0.18 V (TiO 2 ) and −0.14 V (Bi–TiO 2 ) vs Ag/AgCl. Considering that the conduction band (CB) of n-type semiconductors lies ∼0.1 V above the flat-band potential, CB positions were estimated as 0.25 V (TiO 2 ) and −0.08 V (Bi–TiO 2 ) vs NHE. Using EVB = ECB + Eg, valence band (VB) positions were calculated as 2.92 V (TiO 2 ) and 2.06 V (Bi–TiO 2 ). These results indicate that Bi incorporation modifies band alignment, shifting the CB toward a more negative potential and favoring oxygen-related reduction reactions. Having established enhanced visible-light absorption and favorable band alignment, charge separation efficiency was evaluated by transient photocurrent measurements. As illustrated in Fig. 3 F, both materials exhibited stable photocurrent responses over five cycles, indicating structural stability and negligible photocorrosion. Notably, Bi–TiO 2 generated higher photocurrent density, suggesting improved separation and migration of photogenerated carriers. This behavior was further supported by electrochemical impedance spectroscopy (EIS), where in the Nyquist plot ( Fig. 3 G), the Bi–TiO 2 curve intercepted the real impedance axis earlier, resulting in a smaller horizontal arc radius, reflecting decreased charge transfer resistance due to Bi-driven band structure modulation, enhanced defect-assisted conductivity, and reduced charge carrier recombination that collectively facilitate interfacial electron transport, conditions that favor ROS production through reactions of photogenerated electrons with O 2 and holes with H 2 O. To assess the functional consequences of the physicochemical and photoelectrochemical modifications, the photocatalytic activity of the surfaces was evaluated using a methylene blue (MB) degradation assay under visible-light irradiation. Control experiments distinguished photocatalytic effects from dye-related photochemical behavior. Under dark conditions, negligible changes in MB absorbance were observed ( Fig. S4A ), indicating minimal adsorption or catalytic degradation. Similarly, the MB + light (no sample) control exhibited a degradation profile comparable to polished cpTi ( Fig. S4B ), confirming that slight dye reduction arises from the intrinsic photosensitizing behavior of MB. Notably, MB degradation occurred in the presence of all surfaces, including pristine cpTi ( Fig. 3 H), confirming PDT efficiency under LED irradiation. This response is consistent with the PDT mechanism. MB absorbs light within the 500–700 nm range, overlapping with the LED spectrum used in this study, and undergoes singlet-to-triplet state transitions that generate ROS [ 40 ]. These reactive species drive redox reactions leading to chromophore disruption, aromatic ring cleavage, dye decolorization, and reduced absorbance [ 41 ]. Therefore, the minor degradation observed in the MB + light control reflects dye self-photodegradation rather than semiconductor-mediated photocatalysis. Similarly, upon irradiation at bandgap-compatible wavelengths, a photocatalyst generates electron–hole pairs. Photogenerated holes react with hydroxyl species to form •OH radicals, while electrons reduce oxygen to produce O 2 • - radicals, both contributing to dye degradation [ 41 , 42 ]. Accordingly, Bi–TiO 2 exhibited the highest photocatalytic activity, reaching ∼60% after 60 min of irradiation—approximately threefold greater than cpTi—whereas TiO 2 produced only a modest increase ( Fig. 3 H and I). Importantly, the substantially higher degradation observed for Bi–TiO 2 relative to the MB + light control confirms that dye self-photodegradation alone cannot account for the results. The enhanced performance of TiO 2 and Bi–TiO 2 coatings can be primarily attributed to increased hydrophilicity and a higher density of oxygen vacancies (OVs), as indicated by XPS analysis ( Fig. 1 E–G). In addition to defect-mediated effects, the superior activity of Bi–TiO 2 is likely associated with bandgap narrowing and suppressed electron–hole recombination induced by Bi incorporation as previously discussed. Notably, under shorter irradiation times (1 and 5 min), which are clinically relevant, fold-change analysis relative to cpTi (excluding PDT effects) revealed that Bi–TiO 2 outperformed TiO 2 , corroborating the accelerated ROS generation kinetics. To confirm the ability to form ROS, we incubated all experimental surfaces with 1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidine (CMH) as a probe to detect and quantify total ROS levels. CMH specifically reacts with ROS, resulting in the creation of a stable paramagnetic radical (CM•). This radical can be detected through electron paramagnetic resonance, enabling the quantification of ROS formation [ 43 ]. As shown in Fig. 3 I and J, Bi-TiO 2 exhibited intensity signals after light exposure, indicating that the activation of the coating with visible light successfully induced the formation of ROS. Additionally, the intrinsic photocatalytic behavior of the same Bi-TiO 2 surface was previously validated in an independent study using methyl orange as a non-photosensitizing dye, further confirming that the enhanced degradation observed herein is associated with semiconductor-mediated photocatalysis rather than dye-sensitized artifacts [ 16 ]. In contrast, TiO 2 and cpTi showed little to no ROS signal, respectively. No obvious signal was further detected for any experimental surface without light exposure. In contrast, the exposure of Bi–TiO 2 to MB and light (PDT) robustly and significantly increased ROS formation (p = 0.0013), showing almost 2-fold increase compared with the Bi–TiO 2 coating alone ( Fig. 3 I). These findings indicate that MB–mediated PDT, in synergy with Bi-TiO 2 photocatalysis, leads to a pronounced increase in ROS generation. The probable mechanism is that visible light excites Bi-doped TiO 2 , whose smaller band gap facilitates the electron transfer process from the valence band (VB) to the conduction band (CB) of TiO 2 and Ti 3+ sites. This electron transfer leads to the reduction of oxygen (O 2 ) to generate •O 2 − radicals. Furthermore, holes (h + ) in the valence band interact with water (H 2 O), resulting in the formation of OH• radicals. To clarify the relative contribution of individual ROS species under synergistic conditions, EPR spin-trapping analyses targeting specific radicals were performed. In the •OH detection region ( Fig. 3 K), the Bi–TiO 2  + MB group exhibited a broader peak-to-valley separation and a deeper valley amplitude, indicating enhanced hydroxyl radical generation, which suggests that •OH plays a major role in the amplified oxidative response. Importantly, analysis of the 1 O 2 (singlet oxygen) region revealed a pronounced signal enhancement for the Bi–TiO 2  + MB condition ( Fig. 3 L). This finding is particularly relevant because 1 O 2 is the primary ROS generated during classical PDT. Upon visible-light excitation, MB transitions to its triplet state and transfers energy directly to molecular oxygen, producing singlet oxygen via a Type II photodynamic mechanism. The increased 1 O 2 signal therefore confirms effective photosensitizer activation and indicates that the Bi–TiO 2 surface does not hinder, and may even facilitate, PDT photochemistry. Thus, the simultaneous enhancement of •OH and 1 O 2 signals indicates that the synergistic system promotes multiple ROS pathways. While 1 O 2 generation is predominantly governed by MB-mediated photodynamic reactions, •OH radicals arise mainly from semiconductor-driven photocatalytic processes, where photogenerated holes oxidize surface hydroxyl groups or water. Altogether, these results support a dual ROS-generation mechanism in which Bi–TiO 2 photocatalysis and PDT act synergistically under visible-light irradiation. Bi incorporation shifted the CB toward a more negative potential ( Fig. 3 M), improved visible-light responsiveness, charge separation, and interfacial electron transfer, favoring •OH formation, whereas the hydrophilic and defect-rich Bi–TiO 2 surface enhanced methylene blue adsorption and photosensitizer activation, promoting 1 O 2 production ( Fig. 3 N). The convergence of these photocatalytic and photodynamic pathways explains the amplified ROS generation observed in both the MB degradation assay and EPR analyses and, given the central role of ROS in oxidative microbial inactivation, suggests improved antimicrobial performance of the Bi–TiO 2 coating. Encouraged by the synergistic ROS generation observed for Bi–TiO 2 and MB-mediated PDT, we subsequently assessed the in vitro antimicrobial efficacy. The analyses were conducted with and without PDT, using a constant LED light power intensity of 45 mW cm −2 for durations of 1 and 5 min. First, a preliminary screening assay was performed using a monospecies biofilm of Staphylococcus aureus ( S. aureus ), as this bacterium represents a well-established in vitro model for evaluating antibacterial activity on biomaterial surfaces. Under dark conditions, Bi–TiO 2 showed a reduction in S. aureus viability compared with TiO 2 , likely due to surface-related effects and mild contact-mediated antibacterial activity associated with Bi incorporation, corroborating with previous study using a dual-species biofilm model [ 16 ]. Upon visible-light irradiation, a time-dependent response was observed, in which no significant reduction occurred after 1 min, whereas a pronounced antibacterial effect was detected after 5 min ( Fig. S5 ). As a Gram-positive bacterium, S. aureus possesses a thick peptidoglycan layer that may confers higher tolerance to oxidative stress, requiring sustained ROS exposure to overcome cellular defense mechanisms. Thus, cumulative ROS generation over 5 min likely exceeded the antioxidant capacity of the cells, leading to pronounced inactivation. Next, considering that peri-implant mucositis and peri-implantitis are biofilm-driven polymicrobial diseases [ 44 ], we evaluated the antimicrobial performance in a 24 h biofilm model. Polymicrobial biofilms exhibit increased pathogenicity due to microbial succession and are embedded in a protective extracellular matrix (exopolysaccharides, eDNA, and proteins), which enhances resistance and microbial cooperation [ 44 ]. To better reproduce the clinical scenario, a saliva-derived microcosm biofilm was used to assess the antimicrobial activity of the photocatalytic Bi–TiO 2 coating combined with PDT ( Fig. 4 A). Importantly, first, an additional control assay confirmed that neither the alkaline pH of the MB solution nor variations in culture medium pH influenced microbial viability, excluding pH as a confounding factor in the possible observed antimicrobial effects ( Fig. S6 ). Fig. 4 In vitro antimicrobial activity of coatings associated with or not associated with PDT (MB + light) was performed for 1 or 5 min of irradiation. (A) Schematic overview of the in vitro microbiological methodology. (B) Representative LIVE/DEAD images of polymicrobial biofilms after each treatment. (C) Quantification of live (green) and dead (red) bacteria from fluorescence images. (D) Viable bacterial counts (log10 CFU/mL) of polymicrobial biofilm (human saliva inoculum) formed on cpTi, TiO 2 , and Bi–TiO 2 under MB-free conditions (MB−) and after methylene blue exposure (MB+) evaluated in the dark. (E) Representative SEM micrographs showing morphological changes after PDT; white arrows indicate cell shrinkage and membrane disruption consistent with oxidative stress. (F) Viable bacterial counts (log10 CFU/mL) of polymicrobial biofilm under dark and after PDT using visible light for 1 min (MB+ 1 min) or 5 min (MB+ 5 min) (G) Schematic illustration of ROS-mediated antibacterial mechanisms. (H) Bacterial cell membrane permeability determined by o-nitrophenyl-β-d-galactopyranoside (ONPG) hydrolysis per mg of protein. Quantitative data are presented as mean ± SD (n = 6). Statistical comparisons were performed using two-way ANOVA followed by Tukey's post hoc test. In all figures, bars marked with asterisks (∗) indicate significant differences between groups, except in Fig. 4 F. In Fig. 4 F, asterisks (∗) indicate comparisons between light conditions (dark vs light) within the same surface/treatment, and hashes (#) indicate comparisons between surfaces within each PDT time condition (cpTi vs TiO 2 vs Bi-TiO 2 ) (∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001; #P < 0.05). Illustrations created with BioRender (Barao, V.; 2025). Fig. 4 In vitro antimicrobial activity of coatings associated with or not associated with PDT (MB + light) was performed for 1 or 5 min of irradiation. (A) Schematic overview of the in vitro microbiological methodology. (B) Representative LIVE/DEAD images of polymicrobial biofilms after each treatment. (C) Quantification of live (green) and dead (red) bacteria from fluorescence images. (D) Viable bacterial counts (log10 CFU/mL) of polymicrobial biofilm (human saliva inoculum) formed on cpTi, TiO 2 , and Bi–TiO 2 under MB-free conditions (MB−) and after methylene blue exposure (MB+) evaluated in the dark. (E) Representative SEM micrographs showing morphological changes after PDT; white arrows indicate cell shrinkage and membrane disruption consistent with oxidative stress. (F) Viable bacterial counts (log10 CFU/mL) of polymicrobial biofilm under dark and after PDT using visible light for 1 min (MB+ 1 min) or 5 min (MB+ 5 min) (G) Schematic illustration of ROS-mediated antibacterial mechanisms. (H) Bacterial cell membrane permeability determined by o-nitrophenyl-β-d-galactopyranoside (ONPG) hydrolysis per mg of protein. Quantitative data are presented as mean ± SD (n = 6). Statistical comparisons were performed using two-way ANOVA followed by Tukey's post hoc test. In all figures, bars marked with asterisks (∗) indicate significant differences between groups, except in Fig. 4 F. In Fig. 4 F, asterisks (∗) indicate comparisons between light conditions (dark vs light) within the same surface/treatment, and hashes (#) indicate comparisons between surfaces within each PDT time condition (cpTi vs TiO 2 vs Bi-TiO 2 ) (∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001; #P < 0.05). Illustrations created with BioRender (Barao, V.; 2025). Live/dead staining was employed to evaluate the antimicrobial efficacy of Bi-TiO 2 coatings under different conditions. Fig. 4 B shows that PDT-treated cpTi and TiO 2 coatings evoked some red fluorescence (∼20% and ∼28%, respectively) after 1 and 5 min of light exposure, suggesting the antimicrobial effect of PDT alone. The slight increase in red fluorescence for TiO 2 compared to cpTi is likely due to its potential for OV-related photocatalytic activity under visible light exposure. In contrast, Bi-TiO 2 coatings already exhibited ∼20% red fluorescence without light exposure, indicating their intrinsic bactericidal effect ( Fig. 4 C). This suggests that the microbial-killing potency of Bi-TiO 2 is comparable to that of PDT. Interestingly, strong red fluorescence was detected in Bi-TiO 2 upon exposure to light, suggesting that the combined action of Bi-TiO 2 photocatalysis and PDT effectively killed a significant portion of the microorganisms. On the other hand, cpTi controls and TiO 2 coatings evoked prominent green fluorescence in the absence of methylene blue exposure (MB-) as well as with methylene blue (MB+) under dark conditions, which corroborates with quantitative CFU results, in which samples exposed to MB in the dark were similar with those without MB, confirming that MB alone does not exert antibacterial activity ( Fig. 4 D). Next, to investigate the PDT effects, SEM analysis was performed to visualize bacterial morphology on the experimental surfaces after the treatments ( Fig. 4 E). Morphological alterations were observed across the different surfaces when PDT was applied, indicating membrane damage associated with the PDT interaction. While a considerable portion of bacteria on cpTi surfaces remained structurally preserved, the Bi–TiO 2  + PDT groups exhibited pronounced cellular shrinkage and lysis, suggesting a stronger bactericidal effect likely related to enhanced ROS generation. Quantitative CFU analysis corroborated these observations ( Fig. 4 F). Compared with dark conditions, PDT significantly reduced viable colonies after both 1 and 5 min of irradiation. Under dark conditions, Bi–TiO 2 already presented fewer microorganisms than TiO 2 , likely due to the antibacterial contribution of Bi 3+ ions, as reported in our previous work [ 16 ]. In the presence of PDT, cpTi and TiO 2 also showed antimicrobial activity through ROS generated by MB photosensitization. However, Bi–TiO 2 achieved the most pronounced effect, reducing bacterial counts by nearly 3 logs relative to the control. Notably, no significant differences were observed between 1 and 5 min of irradiation, indicating that ROS-mediated killing occurs rapidly, which is clinically relevant because shorter irradiation times are more feasible for practical application. Finally, to better simulate the clinical transition from peri-implant mucositis to peri-implantitis, which is marked by a shift from an aerobic to an anaerobic, pathogen-enriched environment, a validated two-phase 96-h polymicrobial biofilm model was employed. In this model, Bi–TiO 2 combined with PDT significantly reduced biofilm viability after both 1 and 5 min ( Fig. S7A and B ), corroborating the 24-h microcosm findings. In a subsequent recolonization assay, in which treated biofilms were incubated for an additional 24 h to assess regrowth potential, all PDT groups showed reduced recolonization compared with non-PDT controls ( Fig. S7C ). Although no statistically significant differences were observed among surfaces, Bi–TiO 2 demonstrated lower regrowth than cpTi and TiO 2 , suggesting a sustained antimicrobial effect. Building on the results of Bi-TiO 2 coatings above, we assume that the bactericidal effect may occur from the interaction of released Bi 3+ ions with bacterial surfaces, where they bind ionically to glycoproteins in the membrane glycocalyx [ 45 , 46 ]. This process progressively declines bacterial metabolism and weakens the exterior of bacterial components, making cells more susceptible to ROS damage during photocatalysis. In parallel, ROS generated by PDT accumulate intracellularly, further attacking phospholipid membranes and internal components, ultimately leading to bacterial death ( Fig. 4 G) [ [47] , [48] , [49] , [50] ]. Such antibacterial mechanism was further confirmed using the ortho-nitrophenyl-β-galactoside (ONPG) hydrolysis assay, which assesses bacterial membrane permeability [ 51 , 52 ]. Increased ONPG hydrolysis indicates membrane damage that allows substrate entry and cleavage by intracellular β-galactosidase. As shown in Fig. 4 H, PDT groups exhibited higher ONPG hydrolysis than dark controls, indicating singlet oxygen generation induced disruption of membrane integrity. Notably, irradiated Bi–TiO 2 surfaces showed even greater hydrolysis, suggesting pronounced membrane damage when the photocatalytic coating was combined with PDT, likely due to intensified ROS-mediated oxidative stress. While Bi–TiO 2 coatings display remarkable antimicrobial efficacy boosted by PDT, it is essential to confirm their cytocompatibility with both soft tissue and bone cells, as these cell types are directly involved in peri-implant tissue integration and health. To this end, primary human gingival fibroblasts (HGFs) and human bone mesenchymal stem cells (hBMSCs) were cultured on cpTi, TiO 2 , and Bi–TiO 2 surfaces and assessed for viability, metabolic activity, and morphology. Live/dead staining, combined with CCK-8 assays, consistently demonstrated high cell viability and metabolic activity for both HGFs and hBMSCs across all conditions, remaining within the non-cytotoxic range defined by ISO 10993-5 (<30% reduction compared to the untreated control) ( Fig. 5 A–F). In HGFs, Bi–TiO 2 maintained viability despite surface roughness, likely due to biocompatible Bi 3+ ions [ 53 ] and smaller craters mitigating fibroblast stress. At the same time, PDT exposure did not reduce viability, confirming that ROS generated by photocatalysis and PDT did not induce cytotoxic effects ( Fig. 5 A–C). Consistent with the dose-dependent nature of ROS biology, the use of our specific irradiation parameters likely produced transient ROS levels that remain within the antioxidant capacity of mammalian cells, thereby preserving viability despite antimicrobial activation [ 54 , 55 ]. In hBMSCs, metabolism was preserved for up to 3 days of incubation, with no significant differences among surfaces. Live/dead staining also indicated cell proliferation over time, further supporting cytocompatibility ( Fig. 5 D–F). Morphological analyses corroborate these findings. HGFs aligned along cpTi grooves and adopted a three-dimensional arrangement on PEO coatings, with PDT promoting elongation and spatial reorganization, a pattern previously reported under visible-light stimulation ( Fig. 5 A) [ [16] , [56] ]. Similarly, hBMSCs displayed well-spread morphologies with extended cells with filopodia on cpTi, and star-like arrangements with distinct pseudopodia on TiO 2 and Bi–TiO 2 coatings, indicative of strong adhesion to PEO-modified surfaces ( Fig. 5 D). Notably, Bi–TiO 2 showed a modest (∼20%) increase in cell nuclei number compared to cpTi, suggesting a potential to promote osteogenic proliferation through Bi 3+ -mediated bioactivity ( Fig. S8 ) [ 57 ]. Fig. 5 In vitro biological properties of different surfaces and treatments. (A) Representative LIVE/DEAD staining images showing the effects of surfaces under dark and PDT conditions on the viability of human gingival fibroblasts (HGFs) after 1 day of incubation. (B) Quantification of live (green) and dead (red) HGF cells from fluorescence images. (C) HGF metabolic activity under dark and light (PDT) conditions measured by CCK-8 assay. (D) Representative LIVE/DEAD staining images of hBMSCs and cytoskeleton (red) and nuclei (blue) staining showing cell morphology and spreading. (E) Quantification of live (green) and dead (red) hBMSC cells from fluorescence images. (F) hBMSC metabolic activity after 1 and 3 days of incubation measured by CCK-8 assay. (G) Calcium content quantification and representative images of mineralized nodule formation across groups. (H) SEM images and (I) XRD diffractograms of apatite formed on the coatings after 28 days in SBF. (J) Schematic summary of the proposed bioactive effects of Bi–TiO 2 coatings. Quantitative data are presented as mean ± SD (n = 5). Statistical analyses for (B, C, E, and F) were performed using two-way ANOVA followed by Tukey's post hoc test, while (G) was analyzed using one-way ANOVA followed by Tukey's post hoc test (∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001; ns = not significant). Illustration created with BioRender (Barao, V.; 2025). Fig. 5 In vitro biological properties of different surfaces and treatments. (A) Representative LIVE/DEAD staining images showing the effects of surfaces under dark and PDT conditions on the viability of human gingival fibroblasts (HGFs) after 1 day of incubation. (B) Quantification of live (green) and dead (red) HGF cells from fluorescence images. (C) HGF metabolic activity under dark and light (PDT) conditions measured by CCK-8 assay. (D) Representative LIVE/DEAD staining images of hBMSCs and cytoskeleton (red) and nuclei (blue) staining showing cell morphology and spreading. (E) Quantification of live (green) and dead (red) hBMSC cells from fluorescence images. (F) hBMSC metabolic activity after 1 and 3 days of incubation measured by CCK-8 assay. (G) Calcium content quantification and representative images of mineralized nodule formation across groups. (H) SEM images and (I) XRD diffractograms of apatite formed on the coatings after 28 days in SBF. (J) Schematic summary of the proposed bioactive effects of Bi–TiO 2 coatings. Quantitative data are presented as mean ± SD (n = 5). Statistical analyses for (B, C, E, and F) were performed using two-way ANOVA followed by Tukey's post hoc test, while (G) was analyzed using one-way ANOVA followed by Tukey's post hoc test (∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001; ns = not significant). Illustration created with BioRender (Barao, V.; 2025). In addition to being compatible with human cells, it is essential to acknowledge the importance of the bioactivity of novel biomaterials in the osseointegration process. This is particularly significant since the precipitation of calcium and phosphorus ions onto the material surface initiates apatite nucleation, suggesting direct bone bonding. According to a previous review [ 58 ], the bone mineralization initially occurs during the CaP (calcium phosphate) nucleation phase, in which grains are formed within matrix vesicles through a series of enzymatically directed molecular interactions, leading to a localized accumulation of Ca 2+ and PO 4− ions. Subsequently, in the crystal growth phase, preformed apatite crystal templates guide further crystal growth, driven by a continuous supply of Ca 2+ and PO 4− ions from body fluids. Therefore, to assess the bone-bonding potential of the Bi–TiO 2 coating, we used two complementary methods. The first evaluated the ability of cells cultured on the coatings to produce mineralization nodules and calcium-rich deposits, while the second consisted of immersing the materials in simulated body fluid (SBF) for 28 days without cells to determine the coatings’ capacity to induce calcium phosphate precipitation. This acellular CaP formation under SBF conditions serves as an initial indicator of surface bioactivity. Fig. 5 G demonstrates that all surfaces supported mineralization; however, quantitative calcium content was significantly higher for Bi–TiO 2 than for TiO 2 ( p  < 0.05; reported as fold change). Representative digital images corroborate this, showing an apparent increase in mineralization nodule density and stronger staining for Bi–TiO 2 compared with cpTi and TiO 2 . Fig. 5 H displays the apatite layers formed on the surfaces following a 28-day immersion in SBF solution. Notably, the morphology of the apatite also varies depending on the surface type. CpTi exhibited needle-shaped morphology, while TiO 2 and Bi-TiO 2 displayed a spherical morphology covering the entire surface. Consistently, XRD analysis ( Fig. 5 I ) confirmed the nucleation of CaP phases, with characteristic diffraction peaks observed at approximately 25°, 32°, and 46°. According to a previous study [ 59 ], the presence of the rutile crystalline phase and topographical features seems to be a major factor influencing the observed mineralization pattern, crystal orientation and shape of the apatite agglomerates. Such results are significant, as the inflammatory response has been demonstrated to be influenced by the morphology and particle size of hydroxyapatite. Needle-shaped hydroxyapatite may elicite a more prolonged inflammatory response than the spherical-shaped variety found in the PEO coatings. Altogether, these findings indicate that Bi–TiO 2 coatings maintain excellent cytocompatibility with both gingival fibroblasts and bone mesenchymal stem cells, even under light exposure combined with PDT. Moreover, the coating favored early mineralization events, indicating that Bi–TiO 2 surfaces may enhance calcium phosphate deposition through favorable physicochemical interactions, providing a potential approach to control the rate of bone mineralization ( Fig. 5 J). Previous studies have affirmed that macrophage polarization plays a key role in the immune dysregulation associated with peri-implant diseases [ 60 ]. Macrophages can respond to environmental cues by adopting different functional phenotypes. In this context, excessive M1 polarization—potentially triggered by wear particles or ions released from implant materials—is associated with bacterially induced pro-inflammatory responses and tissue destruction, whereas M2 macrophages contribute to inflammation resolution and tissue repair [ 60 , 61 ]. Thus, in this study, macrophage behavior was assessed under three complementary conditions: (i) stimulation with surface extracts obtained after incubating the coatings in culture medium for 3 days, which represents the combined effect of released products; (ii) exposure to Bi 3+ ions at concentrations equivalent to those released from Bi–TiO 2 coatings after 3 days, as determined by ICP analysis, to isolate the effect of soluble species; and (iii) direct culture on cpTi and Bi–TiO 2 surfaces to evaluate surface-mediated effects. Regarding SEM observations in the presence of surface extracts ( Fig. 6 A), macrophages exposed to cpTi and TiO 2 extracts predominantly displayed a rounded morphology, typically associated with a M1-like phenotype. In contrast, cells stimulated with Bi–TiO 2 extracts exhibited a more elongated morphology, commonly linked to M2-like polarization [ 61 ]. This morphological pattern was consistent with the cytokine profile obtained by ELISA, in which although TGF-β levels were similar among groups, macrophages exposed to Bi–TiO 2 extracts secreted significantly lower levels of the pro-inflammatory cytokine TNF-α compared with cpTi and TiO 2 ( Fig. 6 B). These findings suggest that Bi 3+ ions released from Bi–TiO 2 coatings may attenuate pro-inflammatory macrophage activation and promote a more reparative immune profile. Fig. 6 Macrophage response to Bi–TiO 2 surfaces and released species. ( A) SEM images and (B) ELISA quantification of TNF-α and TGF-β secretion of THP-1–derived macrophages cultured for 3 days with extracts from cpTi, TiO 2 , and Bi–TiO 2 . (C) SEM and (D) macrophage cytokine secretion exposed or not (untreated) to Bi 3+ solution. (E) SEM images and (F) ELISA analysis of cytokine secretion of macrophages cultured directly on cpTi and Bi–TiO 2 surfaces for 3 days. Data are presented as mean ± SD (n = 4). For panel B statistical analyses were performed using one-way ANOVA followed by Tukey's post hoc test (∗∗p < 0.01). For panels D and F statistical analyses were performed using student's t-test. Fig. 6 Macrophage response to Bi–TiO 2 surfaces and released species. ( A) SEM images and (B) ELISA quantification of TNF-α and TGF-β secretion of THP-1–derived macrophages cultured for 3 days with extracts from cpTi, TiO 2 , and Bi–TiO 2 . (C) SEM and (D) macrophage cytokine secretion exposed or not (untreated) to Bi 3+ solution. (E) SEM images and (F) ELISA analysis of cytokine secretion of macrophages cultured directly on cpTi and Bi–TiO 2 surfaces for 3 days. Data are presented as mean ± SD (n = 4). For panel B statistical analyses were performed using one-way ANOVA followed by Tukey's post hoc test (∗∗p < 0.01). For panels D and F statistical analyses were performed using student's t-test. To further support this hypothesis, macrophages were directly exposed to Bi 3+ ions at concentrations corresponding to those released from the coatings, as estimated by ICP analysis ( Fig. 6 C and D). High-magnification SEM images revealed thin cytoplasmic protrusions terminating in rounded tips, characteristic of filopodial extensions involved in cell probing and interaction ( Fig. 6 C) [ 61 ], suggesting that exposure to Bi 3+ ions may influence macrophage adhesion dynamics and phenotype. Similarly, macrophages cultured directly on Bi–TiO 2 surfaces displayed enhanced spreading and pseudopod extension with a dense network of thin fibrillar structures extending from the cell edge, likely corresponding to filopodial and lamellipodial extensions involved in active cell spreading and substrate probing [ 62 ], further supporting a surface-mediated immunomodulatory effect of Bi–TiO 2 surfaces towards M2 polarization ( Fig. 6 E). Conversely, cpTi exhibited a rounded morphology with minimal pseudopod extension compatible with M1 phenotype. Consistently, although not statistically significant, ELISA results showed a similar trend, with macrophages on Bi–TiO 2 tending to secrete higher levels of TGF-β and lower levels of TNF-α compared with cpTi ( Fig. 6 F), suggesting that the rough and hydrophilic topography, together with the presence of Bi 3+ on Bi–TiO 2 surfaces, may contribute to modulating macrophage behavior toward a less inflammatory and potentially more regenerative immune profile. Encouraged by in vitro results, we next validated our concept by utilizing a rat subcutaneous infection model to assess the in vivo antimicrobial efficacy of Bi-TiO 2 (w/- PDT) within an infection environment. The dorsal subcutaneous infection model provides a controlled platform to investigate host–material–infection interactions and the therapeutic effect of antimicrobial strategies in vivo while minimizing confounding anatomical and biomechanical variables. More specifically, in this study, to create an infected wound model resembling the diverse microbiota present in peri-implant diseases, discs with experimental surfaces were first subjected to biofilm formation using human saliva as inoculum under microaerophilic conditions, followed by anaerobic incubation [ 63 ]. Then, the biofilms were subjected to mechanical debridement to reduce the robust biofilm, simulating adjunctive mechanical therapy while minimizing the excessive inflammatory burden on the animals. This was followed by treatment, with or without PDT, involving 1 min of irradiation. Subsequently, discs with experimental surfaces and biofilms were implanted in the dorsal subcutaneous region of rats. Both control cpTi controls and Bi-TiO 2 were evaluated under three different treatments: those infected and treated with PDT (methylene blue + 1 min irradiation), those infected but not treated (dark condition), and the uninfected group (sterile condition). After 3 days of implantation, the discs and surrounding tissue were removed for analysis ( Fig. 7 A). Importantly, irradiation was performed prior to implantation to ensure standardized light delivery, since in vivo illumination would be strongly influenced by inter-animal differences in skin thickness, vascularization, and tissue optical properties, which affect light scattering and absorption and could lead to heterogeneous activation of the photosensitizer and coating. This approach enabled a more reliable evaluation of the intrinsic antimicrobial effect of the treatment while avoiding additional surgical manipulation that could introduce inflammation unrelated to the material or therapy. Fig. 7 In vivo assessment of the antibacterial and host-response effects of Bi–TiO 2 versus cpTi using a subcutaneous implantation model (3 days). (A) Experimental design and treatment groups (sterile control, contaminated/dark, contaminated + PDT). (B) Viable bacteria adhered to the discs quantified as log 10  CFU/mL after 3 days. (C) Quantification of inflammatory-cell area density in peri-implant (peridisc) soft tissue. (D) Representative hematoxylin and eosin (H&E) micrographs of tissues surrounding sterile (control), contaminated (dark), and contaminated + PDT samples. (E–F) Quantification of mature (yellowish-red) and immature (greenish-yellow) collagen fibers from Picrosirius Red (PSR) staining under polarized light. (G) Representative PSR images illustrating collagen organization for each condition. (H) Levels of pro-inflammatory cytokines (IL-6, IL-1β, IL-17, TNF-α) in peri-implant tissue. (I) Schematic summary of the proposed in vivo effects of the Bi–TiO 2 coating. For panels (B, C, E, and F), data are presented as mean ± SD (n = 5) and were analyzed using two-way ANOVA followed by Tukey's post hoc test. For panel (H), data are presented as median with interquartile range (n = 5) and were analyzed using Kruskal–Wallis test followed by Dunn's multiple-comparison test. Asterisks (∗) indicate significant differences between surfaces within the same condition, and hashes (#) indicate significant differences between dark and light (PDT) conditions within the same surface group (∗P < 0.05; #P < 0.05). aPDT: antimicrobial photodynamic therapy; PSR: Picrosirius Red. Illustrations created with BioRender (Barao, V.; 2025 ). Fig. 7 In vivo assessment of the antibacterial and host-response effects of Bi–TiO 2 versus cpTi using a subcutaneous implantation model (3 days). (A) Experimental design and treatment groups (sterile control, contaminated/dark, contaminated + PDT). (B) Viable bacteria adhered to the discs quantified as log 10  CFU/mL after 3 days. (C) Quantification of inflammatory-cell area density in peri-implant (peridisc) soft tissue. (D) Representative hematoxylin and eosin (H&E) micrographs of tissues surrounding sterile (control), contaminated (dark), and contaminated + PDT samples. (E–F) Quantification of mature (yellowish-red) and immature (greenish-yellow) collagen fibers from Picrosirius Red (PSR) staining under polarized light. (G) Representative PSR images illustrating collagen organization for each condition. (H) Levels of pro-inflammatory cytokines (IL-6, IL-1β, IL-17, TNF-α) in peri-implant tissue. (I) Schematic summary of the proposed in vivo effects of the Bi–TiO 2 coating. For panels (B, C, E, and F), data are presented as mean ± SD (n = 5) and were analyzed using two-way ANOVA followed by Tukey's post hoc test. For panel (H), data are presented as median with interquartile range (n = 5) and were analyzed using Kruskal–Wallis test followed by Dunn's multiple-comparison test. Asterisks (∗) indicate significant differences between surfaces within the same condition, and hashes (#) indicate significant differences between dark and light (PDT) conditions within the same surface group (∗P < 0.05; #P < 0.05). aPDT: antimicrobial photodynamic therapy; PSR: Picrosirius Red. Illustrations created with BioRender (Barao, V.; 2025 ). To assess the in vivo antimicrobial potential of Bi-TiO 2 , we investigated its bactericidal effects, both with and without PDT treatment, against polymicrobial biofilms using the agar plate dilution method [ 16 ]. As shown in Fig. 7 B, only photocatalytic Bi-TiO 2 demonstrated significant bacterial reduction after PDT treatment compared to the infected dark condition, indicating the boosting effect of ROS from photocatalysis + PDT, which does not occur for cpTi controls. PDT treatment demonstrated that the Bi–TiO 2 coating significantly reduced the number of microbial colonies compared to cpTi controls, resulting in a substantial decrease in polymicrobial biofilm within a very short irradiation time. Tissue samples were obtained from the regions surrounding the infected implant sites in the rat and were subsequently examined histologically to evaluate the extent of inflammation and infection levels. H&E staining showed a substantial infiltration of inflammatory cells, distinguished by their purple staining and spherical morphology, in the presence of the biofilm formed on cpTi (dark condition) ( Fig. 7 C and D). In comparison, the Bi-TiO 2 group exhibited a slight reduction in the presence of inflammatory cells, indicating a less pronounced inflammatory response, which is likely attributed to its bactericidal and bioactive properties, as demonstrated in the in vitro findings. In contrast, following PDT treatment, a noticeable reduction in the quantity of inflammatory cells was observed on both surfaces, and significantly fewer inflammatory cells were found in the tissues surrounding Bi-TiO 2 compared to cpTi. This suggests that the combination therapy of Bi-TiO 2 and PDT effectively alleviates the in vivo inflammatory response. Furthermore, Picrosirius red staining was employed to assess the quantity and type of collagen formation. The results, as seen in Fig. 7 E–G, show a higher percentage of mature (yellowish-red) to immature (greenish-yellow) collagen in all Bi-TiO 2 groups, primarily due to the short-term (3-day) implantation period, where mature collagen may be predominantly derived from pre-existing tissue rather than newly formed collagen during wound repair. The process of collagen synthesis and maturation in wound healing is dynamic, with initial deposition of immature collagen followed by remodeling over weeks to months, resulting in mature collagen with increased strength [ 64 ]. The higher presence of greenish-intensity immature collagen in contaminated samples (dark and PDT) compared to non-contaminated control conditions suggests an intricate interplay between bacterial presence and host tissue response. This pattern remains consistent for PDT-treated samples compared to dark controls, indicating that the new collagen formation may suggest the wound-healing ability of the treated Bi-TiO 2 [ 65 , 66 ]. Further quantification of pro-inflammatory cytokines via multiplex assay validated these results. The study of cytokine patterns in biofilm-related infections is crucial, as bacterial products stimulate various cell types, including monocytes/macrophages, lymphocytes, fibroblasts, and endothelial cells, leading to the secretion of proinflammatory and immunoregulatory cytokines. When bacteria or their products penetrate the tissues, they trigger the recruitment and activation of the monocyte/T lymphocyte axis, ultimately leading to an increased release of pro-inflammatory cytokines associated with osteoclast activity and tissue deterioration [ 66 , 67 ]. According to previous clinical trials, such expression levels of pro-inflammatory cytokines are reduced after PDT treatment of peri-implantitis and periodontitis, owing to its capacity to decrease certain critical virulence factors, including lipopolysaccharide and proteases [ 68 ]. Similarly, although not significant, lower expression levels of pro-inflammatory cytokines IL-6, IL-1b, TNF-a, and IL-17 were found for Bi-TiO 2 , especially after PDT treatment, in which the combined strategy demonstrated the expression of pro-inflammatory-related factors slightly closer to that of sterile conditions ( Fig. 7 H). These results illustrated in Fig. 7 I demonstrate the antimicrobial and anti-infective properties of Bi-TiO 2  + PDT at a more physiological level for biological applications. Although the dorsal subcutaneous infection model provides valuable information regarding the host inflammatory response to infected implants and the therapeutic effect of the proposed treatment, some limitations must be acknowledged. Importantly, the subcutaneous soft tissue setting allowed the assessment of local tissue reaction and inflammatory cell infiltration around an infected implant-like substrate, providing insights into early host–material–infection interactions and their modulation by the PDT treatment. Thus, the results obtained should be interpreted as a controlled evaluation of host–material–infection interactions rather than a direct simulation of peri-implantitis progression in the oral cavity, as this model does not fully reproduce the complex biological and biomechanical environment of the oral cavity (e.g. influence of saliva, masticatory forces, bone remodeling). It is also important to emphasize that this model was not designed to replicate intraoral optical conditions, as the therapeutic concept does not rely on light penetration through intact peri-implant soft tissues. The clinical applicability of this strategy must therefore be considered alongside realistic light-delivery conditions and established PDT workflows. From a translational perspective, the proposed therapy could be clinically implemented following protocols previously reported for antimicrobial photodynamic therapy (aPDT) in peri-implant sites [ 69 , 70 ]. The LED device employed operates within the spectral range widely used for MB-mediated PDT, particularly near the absorption peak of methylene blue (∼660 nm). Although visible light in this range exhibits limited penetration through intact gingival or peri-implant mucosa due to absorption and scattering, the therapeutic concept proposed here is not dependent on trans-mucosal irradiation. Instead, it is primarily intended for clinical scenarios involving open-flap debridement, a standard intervention in peri-implantitis management, where mucoperiosteal flap elevation provides direct access to the implant surface. Thus, the photosensitizer, in this case MB, could be applied topically to the dental implant surface during open-flap surgery or in a non-surgical approach, being introduced into the peri-implant pocket. A pre-irradiation time of approximately 1–3 min is commonly applied to allow adsorption of the photosensitizer onto both the implant surface and the residual biofilm. Subsequently, light irradiation could be delivered using a fiber-optic tip inserted into the peri-implant pocket, following clinically established protocols typically ranging from 10 s to 3 min per site [ 69 , 70 ]. Under these conditions, it is hypothesized that the adsorbed MB would localize on both the implant surface and the biofilm remaining in areas exposed by bone loss, and that localized light delivery through the fiber tip would activate the coating and the photosensitizer simultaneously, promoting a combined photocatalytic and photodynamic antimicrobial effect in situ . Future developments may explore minimally invasive light-delivery strategies, such as the use of thin optical LED fibers inserted into peri-implant pockets, enabling localized irradiation of the implant surface without the need for surgical exposure. Such approaches could expand the clinical applicability of photocatalysis-assisted PDT while maintaining the principles of minimally invasive therapy. In parallel, further investigations using intraosseous implantation models will be necessary to evaluate the long-term osseointegration behavior of Bi–TiO 2 coatings, including parameters such as bone–implant contact and bone volume fraction.

We present herein a proof-of-concept for the design and performance of Bi-TiO 2 coatings regarding multiple aspects relevant to dental implant surfaces. Bi-TiO 2 coatings were found to exhibit a mixed crystalline phase structure, which enhanced their tribological properties. Regarding cellular responses, Bi-TiO 2 coatings did neither compromise the viability of primary cells of the periodontium in vitro nor the soft tissue responses in vivo . Most importantly, the antibacterial efficacy of Bi-TiO 2 coatings showed superiority in an in vitro polymicrobial biofilm model and a subcutaneous infection model. Integrating Bi–TiO 2 photocatalysis with PDT under a single visible-wavelength exposure offers a smart and safe strategy to minimize peri-implant infections. Mechanistically, under visible light the Bi dopant narrowed the TiO 2 band gap and stabilized charge carriers, sustaining photocatalytic generation of ROS, while the hydrophilic Bi–TiO 2 surface was able to adsorbs methylene blue, enabling photodynamic activation under the same wavelength. By coupling strong antimicrobial action with the preservation of key surface attributes such as wear resistance, biocompatibility, and favorable cell–material interactions, Bi-TiO 2 holds great potential for advancing modern implantology and extending implant longevity.

Discs (Φ 10 mm × 1 mm) of grade II commercially pure titanium (cpTi) (Realum Industry and Commerce of Pure Metals and Alloys Ltd) were prepared according to a previously reported method [ 16 , 71 ] to be employed for in vitro and in vivo assays. Briefly, the substrates were standardized by mechanical polishing and an ultrasonic cleaning process before coating deposition. The polished and cleaned Ti discs were used as controls (cpTi). To prepare photocatalytic coatings on discs, two solutions were prepared in deionized water: 0.8 M of ammonium acetate (CH 3 COONH 4 ) (Synth) and 20 mL of nitric acid (65 wt%) dissolved in 480 mL of deionized water were used for the TiO 2 group; Bi-TiO 2 coating was fabricated using the same electrolyte with the addition of 0.001 M of bismuth nitrate (Dinamica Ltd). Both surfaces were synthesized using PEO technology with a pulsed direct current (DC) power supply and a cooling system (Plasma Technology Ltd.). The parameters used were 250 V, 1000 Hz, and a 10% duty cycle for 7 min [ 16 ]. Then, samples were washed several times with deionized water and air-dried. A scanning electron microscope (SEM, JEOL JSM-6010LA) equipped with energy-dispersive spectroscopy (EDS, JEOL JSM-6010LA) was used to observe the surface morphology, elemental composition, and distribution of the coatings. Three-dimensional images of disc topography and its surface area were obtained by confocal scanning laser microscopy (CSLM, VK-X200, Keyence) [ 72 ]. The crystalline composition of the groups was investigated using an X-ray diffractometer (XRD, PANalytical X'Pert 3 Powder) with monochromatic Cu Kα radiation (λ = 1.540598 Å, 2θ = 20–80°) operating at 45 kV and 40 mA. X-ray Photoelectron Spectroscopy (XPS) was carried out with a spectrometer (hemispherical analyzer) (K-Alpha X-ray XPS, Thermo Scientific) to confirm the chemical states of individual elements [ 73 ]. Samples were also characterized for wettability, measured using a water contact angle meter (Ramé-Hart 100-00, Ramé-Hart Instrument Co.), and surface roughness, accessed by a profilometer (Dektak D150, Veeco) [ 74 ]. Bi ion release from coatings was determined by inductively coupled plasma optical emission spectrometry (ICP-OES; iCAP 6000, Thermo Fischer Scientific Inc.) as previously reported [ 75 ]. In summary, sterilized samples were immersed in phosphate-buffered saline (PBS), pH 7.4 (Gibco™, Life Technologies), and maintained under agitation (90 rpm) at 37 °C. Aliquots were collected at 1, 2, 4, 24, 72, and 144 h, diluted in HNO 3 65%, and the Bi release content was measured using ICP-OES [ 75 ]. Samples were tested by a custom-made pin-on-disk tribometer (Faculty of Mechanical Engineering, University of São Paulo) and an indenter (HMV-2 Micro Hardness Tester, Shimadzu Co) to verify the wear resistance of the surfaces and the Vickers hardness (VHS), respectively. The tribological assessment was performed to measure the friction coefficient of surfaces using a counter body of Zr (Y-TZP, Φ 5 mm) against the samples immersed in simulated body fluid solution (SBF) utilizing a vertical load of 5 N, track diameter of 7.6 mm, sliding velocity of 0.01 m s −1 , and sliding duration of 300 s [ 39 ]. Mass loss (μg) from the surface during the tribological test was measured using a precision scale (AUY-UNIBLOC Analytical Balance, Shimadzu Corporation) to weigh samples before (baseline) and after tribological tests. In addition, the wear track was characterized after the tribological tests experiment. SEM images were obtained to verify the morphology of the wear scars and an optical microscope (VMM-100- BT; Walter UHL) equipped with a digital camera (KC-512NT; Kodo BR Eletrônica Ltd.) and an analyzer unit (QC 220-HH Quadra-Check 200; Metronics Inc.) was employed to evaluate the wear track width and length in relation to the total surface area to calculate the total wear area, as previously reported [ 39 ]. For VHS measurement, samples were indented with 0.5 kgf for 15 s in four randomly dispersed areas [ 73 ]. To elucidate the optical, electrochemical, and photoelectrochemical properties of the coatings, a combination of spectroscopic and electrochemical techniques was employed [ 76 ]. UV–vis diffuse reflectance spectroscopy (DRS) was performed using a Cary 5000 UV–vis–NIR spectrophotometer (Agilent, USA) to evaluate the light-absorption behavior of the samples within the wavelength range of 300–850 nm. The reflectance data (R) were converted using the Kubelka–Munk function, and the bandgap energies (Eg) were estimated by the Tauc plot method through linear extrapolation of the absorption edge. Valence band X-ray photoelectron spectroscopy (VB-XPS) measurements were conducted by fixing the samples onto the holder with conductive tape. High-resolution scans were recorded over the binding energy range of −5 to 20 eV, and the valence band maximum (VBM) position was determined by linear extrapolation of the leading edge of the VB spectrum to the baseline, representing the energy separation between the Fermi level and the valence band edge. Transient photocurrent (I–t) responses, electrochemical impedance spectroscopy (EIS), and Mott–Schottky (M − S) analyses were performed using an AUTOLAB electrochemical workstation (Switzerland) in a conventional three-electrode configuration consisting of a platinum counter electrode, an Ag/AgCl reference electrode, and fluorine-doped tin oxide (FTO) conductive glass coated with the sample as the working electrode. A 0.5 M Na 2 SO 4 aqueous solution was used as the electrolyte. Photoelectrochemical measurements were carried out under illumination provided by a 500 W full-spectrum xenon (Xe) lamp. For transient photocurrent measurements, repeated 50-s light on/off cycles were applied. During EIS and Mott–Schottky analyses, the Xe lamp remained continuously on. The temperature of the light source was controlled at approximately 20 °C using a circulating water-cooling system to ensure thermal stability. The generation of •OH, 1 O 2 , and total ROS under visible light irradiation was assessed by an electron spin resonance spectrometer (ESR, EMXplus, Bruker). For total ROS, samples were mixed with 1 mL of 200 μM CMH (1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidine) solution in Krebs-HEPES buffer in a 24-well plate and irradiated for 10 min [ 77 ]. The concentration of total ROS was determined by calculating the area under the curve in the ESR spectra obtained. For selective ROS identification, spin-trapping experiments were performed using TEMP (2,2,6,6-tetramethylpiperidine) and DMPO (5,5-dimethyl-1-pyrroline-N-oxide) as trapping agents for 1 O 2 and •OH, respectively. Briefly, 15 μL of TEMP or 50 μL of DMPO were added to 500 μL of ultrapure water, and samples were immersed in the trapping solution. The formation of radical adducts was analyzed by EPR after 10 min of visible-light irradiation. The decomposition of MB was quantified to evaluate the photocatalytic activity of the samples under both dark and visible-light conditions. Samples were immersed in 2 mL of a 10 mg L −1 MB solution and maintained in the dark for 30 min to establish adsorption–desorption equilibrium (pre-adsorption step). Subsequently, specimens were exposed to visible-light irradiation (λ = 420–690 nm) using a 105 W light-emitting diode (LED) lamp (Ledsim). The emission spectrum of the LED lamp was experimentally recorded using a fiber-optic spectrometer (Ocean Optics spectrometer USB2000, USA), confirming continuous spectral output across the visible range. The spectral distribution, including the dominant emission band centered at ∼560 nm, is presented in Fig. S3 . The distance between the LED source and the sample surface was fixed at 6.5 cm using a customized irradiation apparatus to ensure reproducible geometric alignment. The irradiance reaching the sample (∼45 mW cm −2 ) was measured with a calibrated optical power meter (Ophir Optronics Solution Ltd., Israel). Light exposure was performed for 0–60 min, and MB degradation was quantified by monitoring the absorbance changes of the dye solution. All experiments involving light irradiation were conducted according to the parameters above. Dark-condition controls were conducted by fully wrapping the plate in aluminum foil to prevent light exposure. Photocatalytic activity was assessed through the degradation of MB, indicated by the decrease in its intensity, which was quantified using a spectrophotometer (DU 800 UV–Visible Spectrophotometer, Beckman Coulter Inc.) at 464 nm. The formula for calculating photocatalytic activity (%) is presented in our previous study [ 16 ]. The monospecie biofilm model of S. aureus was used to screening the antimicrobial efficacy of coatings associated with or not associated with the PDT strategy, in both dark and visible light irradiation conditions. To carry out the PDT, an MB solution at a concentration of 200 μg mL −1 was prepared using phosphate buffer at a pH of 10 [ 12 ]. Firstly, three volunteers (Ethics Committee number 19100719.0.0000.5418) provided fresh human saliva that was collected under stimulation immediately before the experiment began. Then, a pool of this saliva was centrifuged, filter-sterilized, and 1 mL of the resultant solution was incubated with sterile samples (UV-light irradiation for 30 min on each side) for 30 min at 37 °C to form a salivary pellicle [ 78 ]. The bacterial strain (ATCC® 25923™) was stored in 20% glycerol at −80 °C until use. For activation, colonies were streaked onto appropriate brain heart infusion (BHI, Difco Laboratories, France) agar plates and incubated at 37 °C for 24 h. Selected colonies were then transferred to BHI broth medium and incubated overnight under the same conditions. Cells were harvested by centrifugation (6000× g , 5 min, 4 °C), washed twice with 0.9% NaCl, and resuspended in BHI broth. The inoculum was adjusted to the desired optical density (OD 550nm  = 0.15; ≈10 6  cells/mL). The disks previously coated with salivary pellicle were placed in 24-well plates containing 900 μL growth medium and 100 μL bacterial suspension and incubated at 37 °C for 24 h to allow biofilm formation. After 24h, samples were washed in NaCl 0.9% and transferred to a new 24-well plate containing 1 mL of PBS for the groups that will not be submitted to the PDT (MB-), and plates with 100 μL of methylene blue (200 μg/mL) for the groups submitted to PDT (MB+). MB + groups were kept in contact with the solution for 1 min in the dark (foil-wrapped plate) to allow the MB to adsorb to the interior of the bacteria [ 79 ]. The duration of visible light irradiation was tested using 1 and 5 min of exposure under the same conditions as described in the photocatalysis experiment, specifically for the groups subjected to PDT. To control the light influence, all groups were also tested in the dark. Subsequently, the discs were washed once in 0.9% NaCl and transferred to cryogenic tubes containing 2 mL of 0.9% NaCl for sonication (7W, 30 s), vortexing, and serial dilution. Next, aliquots of each dilution were plated on BHI agar and incubated for 24 h at 37 °C for bacterial counts (log 10  CFU/mL) [ 74 ]. For the polymicrobial assay, biofilms were formed by incubating salivary pellicle-coated discs with stimulated human saliva (unfiltered) in BHI medium for 24 h at 37 °C and 10% CO 2 to better reproduce the oral microbial community. After biofilm development, the same PDT protocol described for the monospecie model was applied (MB 200 μg/mL, 1 min dark pre-incubation, and 1 or 5 min visible-light irradiation), including dark controls. Biofilms were then collected by sonication, vortexed, serially diluted, and plated on Columbia Blood agar (CBA) for CFU quantification [ 74 ]. In addition, samples from each group that was not sonicated neither vortexed were evaluated by CLSM (LSM 800, Zeiss) to observe the viability of adherent bacteria on each coating after being stained with the LIVE/DEAD BacLight Bacterial Viability Kit (ThermoFisher), wherein mixed SYTO 9 was used to observe viable cells (green fluorescence channel) and propidium iodide dye to evaluate nonviable cells (red fluorescence channel). After scanning the samples, 3D reconstructions were obtained, and the fluorescence percentage of each channel was quantified using the ZEN software (Zeiss) [ 39 , 75 ]. In addition, samples from each group were evaluated using a SEM (JEOL JSM-6010LA) after biofilm formation and application of the treatment protocols to observe changes in biofilm structure. For this, the samples were washed twice, fixed with Karnovsky solution, and serially dehydrated using ethanol washes as described elsewhere [ 80 ]. Then, the discs were air-dried and gold-sputtered for investigation by SEM. Finally, to further simulate the ecological transition associated with peri-implant disease progression, a validated two-phase polymicrobial biofilm model was employed [ 81 ]. In this model, saliva-coated samples were incubated under microaerophilic conditions (37 °C, 10% CO 2 ) in a nutritionally enriched medium (modified fluid universal medium supplemented with 10% BHI and 10% sucrose) to promote the establishment of early colonizers and initial biofilm structuring. In the second phase (48–96 h), samples were transferred to fresh supplemented medium with a new inoculum and incubated under strict anaerobic conditions, favoring the outgrowth of anaerobes. The culture medium was renewed every 24 h. After a total of 96 h, samples were subjected to the treatment protocols, followed by washing, sonication-assisted biofilm detachment, serial dilution, and CFU quantification on CBA under anaerobic incubation (37 °C). For the recolonization assay, additional samples underwent the same 96 h biofilm formation protocol and corresponding treatments. However, instead of immediate processing for CFU determination, treated discs were transferred to wells containing fresh supplemented culture medium and incubated for an additional 24 h under anaerobic conditions to allow potential biofilm regrowth. After this recovery period, samples were washed and processed by sonication, serial dilution, and plating on CBA for CFU quantification. Bacterial membrane permeability on the sample surfaces was evaluated using an ONPG-based β-galactosidase assay (β-Gal Assay Kit, Invitrogen). In this method, if the treatment compromises membrane integrity, ONPG can penetrate the bacterial cytoplasm and be hydrolyzed by intracellular β-galactosidase into o-nitrophenol (ONP), generating a measurable spectrophotometric signal [ 52 ]. For this, microcosm biofilms were first formed on the samples for 24 h and then subjected to treatment with or without light irradiation. After treatment, biofilms were detached by sonication in 1 mL of 0.9% NaCl, and the resulting suspension was collected. An aliquot of 10 μL of the suspension was used for the assay following the manufacturer's instructions, incubated for 4 h at 37 °C, and the absorbance was measured at 420 nm. The obtained values were normalized to the total protein content of each sample, determined using a BCA protein assay kit (Thermo Fisher Scientific). Human gingival fibroblasts (HGFs) cultured in Dulbecco's modified Eagle medium (DMEM; Gibco™, Life Technologies) and Human Bone Mesenchymal Stem Cells (hBMSc) cultured in minimum essential medium Eagle – alpha modification (Alpha MEM; Gibco™, Life Technologies) were used to assess the cytotoxicity of the coatings and/or irradiation treatment. For both cell lines, the medium was supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin/streptomycin (Gibco) [ 70 ]. Firstly, HGFs were seeded on sample surfaces at a density of 1 × 10 4  cells per well for 24 h. Then, samples were irradiated or not (dark) for 10 min before evaluating the cell metabolic activity by the cell counting kit 8 (CCK-8; Abcam plc, China), according to the manufacturer's protocol. HGF cells were also used to verify cell viability using the LIVE/DEAD™ Cell Imaging Kit (Invitrogen, Life Technologies), following the manufacturer's protocol. Images were acquired using a fluorescence microscope (Zeiss AxioImager Z.1, Carl Zeiss Microscopy GmbH), and live and dead cells were counted using the ZEN software (Zeiss) for percentage calculations [ 75 ]. For hBMSC, the same experiments were performed after 1 and 3 days of incubation, under dark conditions, to investigate cell behavior on the coatings for bone regeneration at the early stages of implant rehabilitation. Cytoskeleton/nuclei staining was conducted to observe hBMSCs' adhesion to coatings, as undertaken previously [ 75 ]. Briefly, after 1 day of cells culture on samples surfaces the experiment was carried out as follow: cells were i) washed once in PBS; ii) fixed for 20 min with 4% formaldehyde solution; iii) permeabilized with 0.5% Triton X-100 in PBS (v/v) for 20 min; iv) washed twice with 0.05% Tween-20 in PBS (v/v); v) incubated in blocking buffer containing 1% bovine serum albumin (BSA) with 100 mM glycine (10:1 v/v) for 30 min at room temperature; vii) incubated with Texas Red™-X Phalloidin (1:200 dilution; Thermo Fischer Scientific Inc) during 1 h for cytoskeleton staining; viii) washed with 0.05% Tween-20 in PBS (v/v); and iv) incubated with 5 mg/mL of 4, 6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich) for the nucleus staining (5 min). Then, the samples were analyzed using a fluorescence microscope (Zeiss AxioImager Z.1), and the number of nuclei per mm 2 was calculated using ImageJ software. To evaluate the osteogenic potential of the coatings, both qualitative and quantitative assessments of mineralized matrix formation were performed. First, hBMSCs were cultured on the different surfaces for 21 days in either osteogenic medium (α-MEM supplemented with 10% FBS, 1% penicillin–streptomycin, 10 nM dexamethasone, 10 mM β-glycerophosphate, and 0.1 mM L-ascorbic acid-2-phosphate) or non-osteogenic medium (α-MEM supplemented with 10% FBS and 1% penicillin–streptomycin). For Alizarin Red staining, cells were fixed with 4% paraformaldehyde, stained with a 2% Alizarin Red S solution, and the stained mineral deposits were imaged under a microscope [ 82 ]. Calcium content was quantified by the o-cresolphthalein complexone (OCPC) assay. After staining, substrates were rinsed twice with PBS and calcium extracted overnight with 0.5 M acetic acid. The OCPC working solution was prepared, then mixed with ethanolamine/boric acid buffer at pH 11, 8-hydroxyquinoline, and demi water to form the reaction mixture. For each measurement, 300 μL of reagent was added to 10 μL of sample. A CaCl 2 standard curve was used for calibration. Plates were incubated at room temperature, absorbance was read at 540 nm, and calcium concentrations were calculated from the standard curve [ 83 ]. To assess the effect of the surfaces and released species on macrophage behavior, the human monocytic cell line THP-1 obtained from ATCC was cultured at 37 °C with 5% CO 2 in RPMI 1640 medium (Gibco) supplemented with 10% FBS, 1% penicillin/streptomycin, and 50 ng mL −1 phorbol-12-myristate-13-acetate (PMA; Sigma-Aldrich) to induce differentiation into macrophage-like cells [ 84 ]. Macrophage responses were evaluated under three experimental conditions in order to distinguish the influence of surface properties from that of released species. For direct surface interaction, 1.5 × 10 4  cells per well were seeded onto sterilized cpTi and Bi–TiO 2 discs placed in 24-well plates. To isolate the effect of soluble bismuth species, macrophages cultured on 48-well tissue culture plates were exposed to Bi 3+ ions from a nitrate bismuth solution at the concentration corresponding to the amount released from Bi–TiO 2 coatings after 3 days, as determined by ICP analysis. In addition, extract-based exposure was evaluated by incubating cpTi, TiO 2 , and Bi–TiO 2 discs in culture medium for 3 days to obtain surface extracts, which were subsequently used to stimulate macrophages cultured on 48-well tissue culture plates. After the 3-day incubation at 37 °C and 5% CO 2 , cells were washed with PBS and subjected to DNA quantification, SEM, and cytokine analyses. DNA content was determined using the Quantifluor dsDNA System (Promega Benelux BV) according to the manufacturer's instructions. Briefly, a standard curve was established in a 96-well plate by serial dilution of dsDNA stock solutions (0–2000 ng mL −1 ). Then, 100 μL of each sample and 100 μL of freshly prepared 1 × Quantifluor dye working solution were added in duplicate. After a 5-min incubation at room temperature in the dark, fluorescence was measured at 480/520 nm. To evaluate inflammatory responses, after the 3-day incubation the culture medium from the same samples were collected and stored at −80 °C until further analysis. The concentrations of the pro-inflammatory cytokine TNF-α and the anti-inflammatory cytokine TGF-β in the supernatants were quantified using ELISA kits from ThermoFisher and Sigma Aldrich, respectively, following the manufacturer's instructions. Cytokine secretion values were normalized to DNA content and expressed as pg per ng DNA. Samples were immersed in a simulated body fluid (SBF) for 28 days to verify the precipitation of apatite-like/hydroxyapatite (HAp). The experiment was conducted according to established protocols described elsewhere [ 59 , 85 ]. Discs were soaked in 10 mL of SBF per cm 2 of surface, which was previously calculated using CLSM surface area analysis. Samples were maintained at 37 ± 1 °C, and the solution was refreshed every 24 h. The HAp formation and morphology were observed using SEM, and its crystalline phases were further confirmed by XRD analysis performed under the same equipment settings previously described [ 86 ]. This experiment aimed to further investigate the antibacterial capacity of the Bi-TiO 2 coating, as well as the PDT strategy, in an in vivo subcutaneous infection model. For this, the animal welfare guidelines (ISO 10993-2:2022) and the Animal Arrive 2.0 [ 87 ] were followed to conduct all the animal experiments. Additionally, this study was approved by the Ethics Committee of Araçatuba Dental School, São Paulo State University (registration number 142-2023). Male Wistar rats (weighing »350 g) were kept in cages in a controlled environment (22 ± 2 °C, 12-h light cycle) during the experiment, where they were fed with solid food. They had unrestricted access to water, except for the 12 h before the surgery. For the experiments, all animals were randomly divided into groups by drawing lots and then randomly distributed by a third party. The infection model was developed using UV-sterilized cpTi and Bi-TiO 2 discs, on which a polymicrobial biofilm was grown to simulate peri-implant diseases, mimicking the dynamic shift from an aerobic to an anaerobic community profile [ 63 ]. Briefly, the freshly stimulated human saliva was collected and adjusted to an optical density of 0.1 at 550 nm to prepare the microbial inoculum. Then, saliva-coated samples were incubated in microaerophilic (37 °C and 10% CO 2 ) in fluid universal medium supplemented with 67 mmol L −1 Sorensen's buffer (“modified fluid universal medium”, mFUM; pH = 7.2) with 10% microbial inoculum, 10% BHI, and 10% sucrose. After 48 h, the discs were washed in 0.9% NaCl and placed in fresh supplemented culture medium containing the inoculum to be incubated in anaerobiosis for an additional 48 h [ 60 , 77 ]. [ 63 , 88 ]Following 96 h of polymicrobial biofilm growth, samples were submitted to mechanical debridement using a titanium curette to replicate clinical adjunct therapy. After this, samples were submitted to the PDT protocol as described in the in vitro antibacterial test, in which discs were exposed to irradiation or kept in the dark for 1 min after exposure to methylene blue. Noninfected (sterile control) discs were used as controls. Irradiation was performed prior to implantation to ensure standardized and reproducible light delivery across samples, avoiding variability in light penetration associated with differences in tissue optical properties among animals. For implantation, the animals were anesthetized through intramuscular injection of xylazine hydrochloride (5 mg kg −1 , Coopers Ltd.) associated with ketamine hydrochloride (50 mg kg −1 , Saúde Animal Ltd.) [ 89 ]. Then, the dorsal hairs of rats were shaved and cleaned with 10% povidone-iodine to expose the area for implantation. CpTi and Bi-TiO 2 discs submitted to 3 different conditions (PDT, dark, and sterile conditions) were randomly implanted into small incisions in the subcutaneous tissue of a rat with minimal tissue damage, and the wounds were sutured. The animals were euthanized after 3 days, and the discs and surrounding soft tissues were collected aseptically [ 89 ] for CFU counts, histological analyses, and cytokine expression. The explanted discs underwent a series of procedures to evaluate the in vivo antimicrobial efficacy of each treatment. Samples were washed, sonicated, vortexed, and serially diluted (as previously described in the in vitro experiment) to be plated on CBA and subjected to anaerobic incubation at 37 °C for subsequent quantification of the viable bacteria adhered to the discs (CFU) [ 16 ]. Histological examination was performed in the subcutaneous tissue surrounding the discs, which were prepared for analysis as stated elsewhere [ 89 ]. To summarize, the subcutaneous tissues were fixed in 4% paraformaldehyde and then embedded in paraffin. Then, samples were serially sectioned (5-μm thick) and stained using Hematoxylin and Eosin (H&E) and Picrosirius red (PSR). The examination was performed using an optical microscope (DM4000B, Leica, x400). For PSR, a polarization contrast feature was employed to differentiate the immature, thin collagen fibers (greenish-yellow color) from the mature, thick ones (yellowish-red color). ImageJ software was used to calculate the density of inflammatory cells in the H&E samples, while QWin software (Leica) was used to quantify the intensity percentage for each color. The subcutaneous tissues were also evaluated using a Multiplex Kit (Luminex, Merck) according to the manufacturer's instructions. The expression of immunoregulatory cytokines, including pro-inflammatory cytokines (IL-6, IL-1, TNF-a, and IL-17), was investigated after a series of preparation steps [ 89 ]. In essence, the subcutaneous tissues were immersed in RNAlater® solution (QIAGEN GmbH) and frozen at −80 °C. The tissue was homogenized in PBS and centrifuged to collect the supernatant for analysis. Protein concentration was measured using the Bradford assay method (BioRad). All quantitative data were analyzed using GraphPad Prism (GraphPad Software, San Diego, CA, USA). The normality of data distribution was initially assessed using the Shapiro–Wilk test. Homogeneity of variances was verified using Levene's test. When both assumptions were satisfied, parametric tests were applied; otherwise, equivalent non-parametric analyses were used. For single-factor comparisons involving more than two groups, one-way ANOVA followed by Tukey's post hoc test was used, whereas comparisons between two groups were performed using an unpaired Student's t-test. For single-factor comparisons, one-way ANOVA followed by Tukey's post hoc test was used. For datasets that did not meet parametric assumptions, a non-parametric Kruskal–Wallis test followed by Dunn's multiple-comparisons post hoc test was applied. These results are presented as median and interquartile range, whereas all other parametric data are expressed as mean ± standard deviation (SD). The significance level was set at α = 0.05 for all analyses. Adjusted p-values from the corresponding post hoc tests were used to determine pairwise differences. Symbols indicating statistical significance are specified in each figure legend. The sample size for each experiment was determined based on prior studies with similar experimental designs and outcomes, ensuring adequate statistical power. A post hoc power analysis indicated that the study design provided a statistical power (1 − β) of at least 0.80 for the primary quantitative outcomes.

In the field of dental implantology, commercially pure titanium (cpTi) and its alloys remain the preferred biomaterials for manufacturing implants due to their excellent mechanical properties and biocompatibility with oral tissues [ 1 ]. Despite the generally high success rates of dental implant procedures [ 2 , 3 ], the rise of peri-implant diseases linked to biofilm accumulation has become a significant clinical concern. These diseases, notably peri-implantitis, affect roughly 20% of patients [ 4 ], leading to a considerable healthcare burden. At the same time, the field is contending with the economic impact of these diseases, in which implant loss accounts for approximately half of the costs associated with dental implant treatment [ 5 ]. Consequently, current efforts in peri-implant infection management focus on controlling and decontaminating biofilms that form on dental implant surfaces. The formation of polymicrobial biofilms on dental implant surfaces is recognized as the primary cause of the development of inflammatory conditions, referred to as peri-implant mucositis and peri-implantitis [ 6 ]. Progressive biofilm maturation and polymicrobial community assembly often result in increased virulence[ [7] , [8] , [9] ], and this process has been implicated as a critical factor in the pathogenesis of biofilm infections [ 9 ]. Such biofilms are challenging to treat because microorganisms are protected by extracellular polymeric substances within well-organized microbial communities enmeshed in a protective three-dimensional matrix [ 10 ]. This phenomenon likely accounts for the absence of consensus regarding optimal treatment approaches for infections associated with dental implants [ 11 ]. Photodynamic therapy (PDT) is rapidly emerging as an effective approach among current treatments that combat biofilm-related infections. In contrast to traditional methods such as mechanical debridement and antibiotic therapy, PDT has gained recognition as a non-invasive antimicrobial technique capable of eliminating peri-implant biofilm without imposing physical harm to the implant material or promoting bacterial resistance, which can occur with the use of curettes and antibiotics, respectively. This therapeutic approach involves exposing the infected area to a photosensitizer and applying light at a wavelength compatible with its absorption spectrum [ 12 ]. In view of the carcinogenic nature of ultraviolet light, clinical applications have predominantly utilized biocompatible synthetic dyes that absorb safe visible light around 660 nm, such as methylene blue. Once irradiated, the photosensitizer generates reactive oxygen species (ROS) that are capable of causing physical damage to bacterial structures and ultimately leading to bacterial cell death [ 12 , 13 ]. Although PDT has been considered a promising option for peri-implantitis management, the lack of clinical consensus and limited understanding of its performance emphasize the need to develop novel strategies that can complement or enhance its therapeutic potential [ 14 , 15 ]. In this context, we have recently proposed an innovative strategy for controlling bacterial infections comprising a photocatalytic coating on titanium synthesized through plasma electrolytic oxidation (PEO) [ 16 ]. Photocatalysis is a phenomenon in which, when the coating is photoactivated at specific wavelengths, it also locally releases ROS. Initially used in environmental engineering for disinfection, photocatalysis is now being explored as a method to combat implant infections. These photocatalytic coatings on implant devices have the potential to efficiently target various bacterial strains without generating unwanted side effects [ 17 ]. In practical applications, particularly in cases of peri-implantitis, light can be applied as an external stimulus to enhance biofilm eradication. However, creating clinically suitable antimicrobial photocatalytic coatings from semiconductors like titanium dioxide (TiO 2 ) requires modifications, such as co-doping with metal ions/anions, like nitrogen and bismuth (Bi), to match visible light wavelengths [ 18 ]. Among TiO 2 dopants, Bi incorporation has been reported to promote bandgap narrowing and enhance visible-light absorption primarily through band-structure modulation rather than plasmonic effects [ 19 ]. Importantly, Bi exhibits low toxicity and favorable biocompatibility, making it particularly attractive for implant-related applications [ 20 ]. In our previous work [ 16 ], TiO 2 coatings doped with Bi were synthesized via PEO and demonstrated promising antibacterial photocatalytic activity under stimulation with visible light. These coatings killed early colonizers, including Streptococcus sanguinis and Actinomyces naeslundii , within 30 min of irradiation. Evidently, transitioning from experimental success to clinical use faces challenges, including effectively disrupting polymicrobial biofilms and defining light parameters that are both efficient and time-feasible. Moreover, implant design should consider both the therapeutic role of light in peri-implantitis treatment, and the preventive attributes required for long-term success, including low cytocompatibility, resistance to bacterial adhesion, mechanical and wear resistance, and the avoidance of pro-inflammatory metallic debris release. In view of this, an integrated and multifunctional coating strategy combining visible-light-driven photocatalytic activity and PDT is proposed herein to simultaneously face titanium implant infections and optimize the implant biointerface for biomedical and dental implant applications. Such integration is possible since both the developed coating and the photosensitizer methylene blue absorbs visible light at compatible wavelengths, thereby enabling their synergistic use in photocatalysis and PDT. As shown in Scheme 1 A, a TiO 2 coating doped with bismuth (Bi-TiO 2 ) was synthesized through PEO. This technology was selected to create a coating with a reduced band gap by incorporating Bi into the electrolyte, resulting in improved physicochemical properties and photocatalytic activity in the visible light spectrum. During the microdischarges of PEO, a crystalline structure was formed that improves hardness, wear resistance, and bioactivity by facilitating apatite formation. Additionally, the Bi-TiO 2 coating demonstrated cytocompatibility with human cells and effectively generated ROS upon visible light irradiation. Moreover, methylene blue-mediated PDT further boosted ROS production, facilitating biofilm penetration and eradication, even in an in vivo environment. This approach attenuated the inflammatory response caused by infection, offering a promising avenue for successful implant rehabilitation ( Scheme 1 B). Scheme 1 Fabrication and effects of the experimental coating. A) Schematic illustration showing the PEO procedure for the fabrication of Bi-TiO 2 coatings. B) Illustrative summary showing visible-light activation of Bi–TiO 2 , generating ROS via photocatalysis, and methylene-blue–mediated PDT, further boosting ROS to enable biofilm penetration and microbial killling, ultimately promoting wound healing. The integrated physical photocatalytic and photodynamic mechanism is depicted in the figure. Illustrations created in BioRender. Barao, V. (2025). Scheme 1 Fabrication and effects of the experimental coating. A) Schematic illustration showing the PEO procedure for the fabrication of Bi-TiO 2 coatings. B) Illustrative summary showing visible-light activation of Bi–TiO 2 , generating ROS via photocatalysis, and methylene-blue–mediated PDT, further boosting ROS to enable biofilm penetration and microbial killling, ultimately promoting wound healing. The integrated physical photocatalytic and photodynamic mechanism is depicted in the figure. Illustrations created in BioRender. Barao, V. (2025).

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data will be made available from the corresponding author on reasonable request.