Dual-function thermoresponsive antibiotic-loaded hydrogel with antimicrobial and osteogenic properties for implant-related infection control

Abstract

Implant-related infections demand advanced biomaterials capable of delivering localized, sustained antimicrobial activity while supporting tissue repair. Here, we present an engineered thermoresponsive hydrogel based on poly(N-vinylcaprolactam) (PNVCL), designed as a dual-function platform that enables temperature-triggered gelation, controlled tetracycline release, and osteogenic support. The PNVCL network was tailored to provide rapid sol–gel transition at physiological temperature, strong adhesion to moist surfaces, injectability into narrow implant geometries, high swelling capacity, slow degradation, and a stable drug-release profile—properties rarely combined within a single hydrogel system. To evaluate material performance, we employed a tiered biological framework comprising cytocompatibility assays, controlled physicochemical analyses, and antimicrobial testing against complex, polymicrobial biofilms, complemented by human in situ and in vivo multispecies infection models. The PNVCL–tetracycline hydrogel demonstrated potent antibacterial activity, preserved human gingival fibroblast viability, and sustained its structural integrity and release characteristics even under infection-associated inflammatory conditions. In vivo, the material simultaneously reduced bacterial burden, modulated pathogenic community structure, and promoted new bone formation with a higher degree of maturation, confirming its dual antimicrobial and osteogenic behavior. By integrating precisely engineered thermoresponsive behavior, controlled drug delivery, robust mechanical and interfacial properties, and validated biological functionality, this PNVCL–tetracycline hydrogel represents a material-driven, clinically relevant solution for treating implant-related infections.

Graphical abstract

Highlights

• •Implant infections are common conditions that can lead to treatment failure.
• •Biomedical engineering has focused on sustained drug-releasing strategies.
• •A tetracycline-loaded hydrogel was developed for peri-implant infection control.
• •The hydrogel showed controlled release, antimicrobial activity, and cytocompatibility.
• •It reduced polymicrobial biofilms, altered their composition, and improved bone repair.

1. Introduction

Oral rehabilitation with dental implants is a widely used and highly predictable therapeutic option for replacing missing teeth [[1], [2], [3]]. However, the adhesion of microorganisms and the accumulation of polymicrobial biofilms with complex composition and architecture may lead to the development of infections associated with dental implants [[4], [5], [6], [7]]. Thus, biofilms on the implant surface trigger host immune responses, leading to inflammatory conditions in the peri-implant tissues. This process manifests initially as mucositis, characterized by inflammation of the peri-implant soft tissues, and can progress to peri-implantitis, which involves inflammatory destruction accompanied by progressive bone loss [[8], [9], [10]]. Approximately 1/3 of the patients and 1/5 of all implants experience peri‐implantitis [10,11], a condition lacking a standardized treatment that often leads to failure and increases costs for both patients and healthcare systems. In addition, the aging of the population has raised the need for rehabilitation with dental implants, which are at risk of microbial-induced disease development [12]. Most clinical treatment approaches focus on the mechanical removal of biofilm structures, combined with adjunctive antimicrobial therapies, which still do not demonstrate long-term effectiveness. Consequently, peri-implantitis remains a highly prevalent disease without a consensus on effective treatments [13]. Therefore, since the treatment of these diseases remains a challenge in dentistry due to the difficulty in removing the complex and pathogenic biofilm [[14], [15], [16], [17]], the development of new therapeutic bioengineered strategies is essential for the prevention, control, and treatment of implant-related diseases [18].

Within the context of natural and synthetic polymers, hydrogels have been widely used in the biotechnology field as drug carriers for the controlled release of drugs. They represent a powerful alternative to prevent the biofilm formed on implant devices [[19], [20], [21], [22], [23]]. Additionally, injectable and stimulus-responsive hydrogels with in situ crosslinking allow this type of biomaterial to flow in the sol-gel state to hard-to-reach sites with minimal tissue gaps in peri-implant diseases, and their self-assembly to specific stimuli can promote their gelation, guiding the drug release [22]. Among stimuli-responsive polymers, poly(N-vinylcaprolactam) (PNVCL) is a thermoresponsive polymer with a low critical solution temperature (LCST) that undergoes a liquid-to-gel transition at approximately 32 °C, which is close to the physiological temperature of the human body [24]. In an aqueous solution, the polymer chains expand due to hydrophilic interactions with water. However, when the system is heated, the polymer chains aggregate and collapse, initiating the gelation process [25,26]. In addition, this polymer has been associated with adequate cell viability in vitro, with the absence of toxic components even after the polymer dissociation process [27,28]. PNVCL has demonstrated positive results in previous in vitro studies as a drug carrier for therapeutic agents, including insulin [29], chitosan [30], and the antibiotic doxorubicin [31]. Therefore, PNVCL is an outstanding bioengineered material, not yet widely explored in the field of implant-related diseases, and it may act as a carrier of antimicrobial agents, such as antibiotics, to promote microbial killing, especially considering the limited efficacy and adverse effects of systemic antibiotic use.

Among the antibiotics used for treating dental implant disease, tetracycline (TC) has been employed for decontaminating the surface of dental implants due to its inhibition of protein synthesis and blocking of the binding between tRNA and mRNA in bacterial cells [[32], [33], [34]]. The broad spectrum of action of this antibiotic, including bacteria associated with peri-implant infections, has supported its application both in coatings for the surface of dental implants and components [35,36], as well as for local application in peri-implantitis [33,37]. Its local application has shown a tendency to reduce peri-implant pockets, bleeding, and anaerobic bacteria counts [37]. However, the rapid release of the drug in implant coatings and the absence of drug substantivity after local application restricts its use for late biological complications, after implant installation. In this sense, the use of a hydrogel loaded with TC, which exhibits adhesive properties on the surfaces of dental implants, could provide greater stability of the biomaterial and controlled drug release at the disease site, making it a promising treatment for peri-implant infections. This approach fosters the antimicrobial effect and promotes the healing process.

Herein, we designed and synthesized a thermoresponsive PNVCL hydrogel via free radical polymerization and subsequently loaded with the antibiotic tetracycline (TC) for potential application in the treatment of peri-implant infections (Fig. 1). We hypothesized that the application of the PNVCL hydrogel in its sol-gel state would allow its flow into areas that are difficult to access for implant surface decontamination, such as thread regions. After its gelation in response to the temperature stimulus at the peri-implant site, the hydrogel should remain stable, ensuring a sustained release of TC. Characterization tests revealed that PNVCL hydrogel loaded with TC exhibits adequate physicochemical properties for application in peri-implant diseases, as it can be easily injected through a narrow needle to reach difficult areas of the implant surface and displays low degradation, thereby allowing for greater stability at the application site. Furthermore, the hydrogel exhibits significant antibacterial activity against in vitro polymicrobial biofilms and elicits a suitable response for the metabolic activity of human gingival fibroblast cells. In biofilms formed in situ and in vivo, the hydrogel significantly reduces bacterial counts and modulates bacterial community composition. Additionally, this hydrogel loaded with TC promotes new bone formation in vivo in areas of contaminated bone defects. Therefore, this work represents a crucial step toward the development of a non-invasive, material-based approach with both antibacterial and osteogenic properties for controlling and treating dental implant infections.

Fig. 1.

Schematic representation of the newly developed hydrogel based on poly(N-vinylcaprolactam) (PNVCL) for the controlled release of tetracycline (TC) for the prevention and treatment of peri-implant infections. The main features, potential clinical applications, and possible biological mechanisms of the hydrogel were outlined (image created using Biorender®, license number VC28DMC0OS).

2. Results and discussion

2.1. Synthesis and characterization of PNVCL hydrogel

To test our hypothesis, NVCL (N-vinylcaprolactam) was employed due to its proven biocompatibility and favorable transition at physiological temperatures [25,26]. PNVCL was synthesized via free radical polymerization (Fig. 2A). TC was loaded into the PNVCL hydrogel (Fig. 2B) at a concentration of 5 mg/mL established by the minimum bacterial concentration (MBC) test, as well as at 5 × and 10 × (25 and 50 mg/mL) the MBC concentration (Table S1, Supporting Information). The morphology of the hydrogel was investigated by scanning electron microscopy (SEM) (Fig. 2C). The cross-sectional SEM micrograph shows an interconnected three-dimensional porous structure. The pores may result from ice crystals forming within the hydrogel during freeze-drying [38]. This interconnected, porous structure is a crucial factor in tissue engineering applications, as it facilitates cell proliferation and the transport of oxygen and nutrients [38]. The resulting hydrogel exhibited a clear appearance, with gelation occurring above its LCST (Fig. 2D). The Fourier-transform infrared spectroscopy (FTIR) spectra of NVCL and its polymerized form (PNVCL) confirmed the successful polymerization of the monomer. In the FTIR spectrum of NVCL, characteristic absorption bands were observed at 3108 cm⁻¹, corresponding to vinyl = C–H stretching vibrations, and at 2993 cm⁻¹ and 2847 cm⁻¹, attributed to aliphatic C–H stretching of –CH₃ and –CH₂ groups (Fig. 2E). The bands at 1648 cm⁻¹ and 1620 cm⁻¹ are associated with the carbonyl (C=O stretching) and the C=C stretching of the vinyl group, respectively. Additional bands at 1477 cm⁻¹ and 1440 cm⁻¹ correspond to C–N stretching and aliphatic –CH₂– bending vibrations. The absorption at 996 cm⁻¹, characteristic of out-of-plane bending of vinyl CH₂ groups, further confirms the presence of the vinyl functionality in the monomer. After polymerization, the FTIR spectrum of PNVCL exhibited significant changes. The disappearance of the characteristic vinyl bands at 3108 cm⁻¹ and 996 cm⁻¹ indicates the consumption of the C=C double bonds, confirming successful polymerization [39,40]. The persistence of the band at 1620 cm⁻¹ is attributed to the carbonyl (C=O) stretching vibration of the lactam ring, while the bands at 1477 cm⁻¹ and 1419 cm⁻¹ are assigned to C–N stretching and aliphatic –CH₂– vibrations of the polymer backbone. Additionally, the appearance of bands at 1195 cm⁻¹, 973 cm⁻¹, and 841 cm⁻¹ is consistent with C–N stretching and skeletal vibrations of the polymer chain, further supporting the formation of PNVCL. Overall, the disappearance of vinyl-related bands and the retention of characteristic amide and aliphatic vibrations in the PNVCL spectrum corroborate the successful polymerization of NVCL into PNVCL. About P407, used as a commercial thermoresponsive hydrogel control, the characteristic bands 2881 cm⁻¹ of C−H bands, 1342 cm⁻¹ of −CH bands, and 1097 cm⁻¹ related to the ether C−O stretching were identified [41].

Fig. 2.

Characterizations of PNVCL. A) Representation of the radical polymerization of PNVCL. B) An illustration of the PNVCL hydrogel loaded with tetracycline (TC) in its non-swollen and swollen state. C) SEM micrograph of PNVCL hydrogel after lyophilization (n = 3). D) Images of the transition of hydrogels at 20 °C and 37 °C (n = 2). E) FTIR spectra of P407, NVCL, and PNVCL with and without TC. F) ¹H NMR spectroscopy of P407, NVCL, and PNVCL with and without TC. G) Rheological properties of hydrogels: storage modulus (G′) and loss modulus (G″) (n = 3). H) Size exclusion chromatography of PNVCL (n = 2).

The polymerization process was thoroughly evaluated using ¹H NMR spectroscopy (Fig. 2F) [42,43]. The spectrum of the monomer NCVL reveals doublet signals at 5.8 ppm and 5.7 ppm, corresponding to the protons present in the vinyl group. The chemical shift at 1.7 ppm is attributed to the CH₂ protons from the caprolactone ring. In comparison, the CH₂ groups adjacent to the carbonyl (C=O) and nitrogen (N) atoms are observed at 2.6 and 3.5 ppm, respectively. Notably, after polymerization, the vinyl peaks are eliminated, resulting in significantly broader chemical shifts. The low concentration of TC enables observation of hydrogen chemical shifts in the aliphatic region, specifically at 2.7 and 3 ppm. The commercial P407 sample was characterized by a triblock structure consisting of a central hydrophobic poly(propylene glycol) (PPG) block flanked by two hydrophilic poly(ethylene glycol) (PEG) blocks. The ¹H NMR spectrum confirmed this structure through the following key signals: a doublet at 1.1 ppm corresponding to the methyl protons (-CH₃) of the PPG units; a strong singlet at 3.65 ppm assigned to the methylene protons (-CH₂CH₂O-) of the PEG chains; and a broad multiplet between 3.3 and 3.8 ppm, representing the overlapping signals of the methylene (-CH₂) and methine (-CH-) protons from both the PEG and PPG backbones. In addition, differential scanning calorimetry (DSC) was performed, showing that the addition of TC to the PNVCL matrix significantly alters the thermal properties of the polymer (Fig. S1) [43,44]. As the drug loading increases, the melting temperature, which is around 100 °C, decreases. The enthalpy measurements start at 48 J/g for the pure PNVCL and rise to 71 J/g at a concentration of 5 mg/mL, ultimately reaching 85 J/g at 50 mg/mL. This indicates that TC interferes with the regular folding of the PNVCL chains, preventing the formation of a well-defined structure. A similar behavior was observed in the glass transition region, around 196 °C, where the presence of TC increases the flexibility of the polymer chains, resulting in a lower Tg. This finding corroborates the results of the rheological experiments.

To evaluate the critical solution temperature, the AR2000 Advanced Rheometer (TA Instruments, Asse, Belgium) was used. The transition temperature of the PNVCL was approximately 34 °C (Fig. 2G), consistent with previous studies in which PNVCL is synthesized in aqueous solution [26,39,45]. In addition, rheological analysis of hydrogels containing varying concentrations of TC revealed that higher TC content corresponded to a reduced sol–gel transition temperature. One hypothesis to explain this finding is that higher antibiotic concentrations, when diffused through the hydrogel matrix, can affect the transition process by increasing molecular weight as the antibiotic binds to the polymer chain [46]. Indeed, a previous study showed that higher molecular weight is associated with lower transition temperatures of PNVCL hydrogels [45]. The time required for the hydrogel to transition was also evaluated, and changes in rheological properties over time were assessed using a time-sweep protocol. Across all samples, the transition occurred within seconds. A longer transition time was observed as TC concentration increased; however, the transition time remained adequate for the intended application, which was less than 30 s (Table 1). The presence of TC may explain this result, as it is dispersed in the hydrogel matrix. At higher concentrations, it may have interfered with the polymer chains’ aggregation, requiring a longer time for them to aggregate and collapse. The chemical shift variations observed in the ¹H NMR spectroscopy indicate changes in the local chemical environment of the polymer hydrogens, suggesting that the presence of TC influences polymer chain organization through non-covalent interactions, such as hydrogen bonding or dipole–dipole interactions. While these results do not constitute direct molecular-level proof, when considered together with the concentration-dependent shifts in sol–gel transition temperature, DSC, and rheological data, they provide consistent and complementary evidence that TC modulates polymer chain mobility, thereby influencing gelation behavior. Regarding the molecular weight assessed by size exclusion chromatography (Fig. 2H), a molecular weight of 30.7 × 10³ g/mol was observed for PNVCL after synthesis, which is similar to the values reported in a previous study for this polymer after the polymerization process [45].

Table 1.

In vitro characterization of P407 and PNVCL hydrogels with different TC concentrations (n = 3): T_sol-gel Sol-Gel Transition Temperature, Storage (G′) and Loss (G″) Moduli, Tan δ, and gelation time (s).

Groups	T (°C), Mean ± SD	G′ (Pa), Mean ± SD	G′’ (Pa), Mean ± SD	Tan δ, Mean ± SD	Gelation time (s)
P407	17.00 ± 0.01 (D)	23.47 ± 5.07 (C)	9.87 ± 4.32 (D)	0.4 ± 0.09 (B)	6.35 ± 0.01 (C)
PNVCL	34.00 ± 0.01 (A)	105.40 ± 3.87 (B)	97.28 ± 2.54 (C)	0.92 ± 0.06 (A)	6.36 ± 0.06 (C)
PNVCL/TC 5 mg/mL	32.01 ± 0.01 (B)	128.81 ± 21.77 (B)	117.37 ± 9.81 (B)	0.93 ± 0.08 (A)	19.39 ± 0.06 (B)
PNVCL/TC 25 mg/mL	31.00 ± 0.01 (C)	141.93 ± 5.02 (B)	135.44 ± 3.41 (B)	0.95 ± 0.01 (A)	25.86 ± 0.03 (A)
PNVCL/TC 50 mg/mL	30.99 ± 0.01 (C)	234.16 ± 5.94 (A)	191.52 ± 2.59 (A)	0.82 ± 0.01 (A)	25.91 ± 0.10 (A)
Different letters indicate statistically significant differences among the groups (p < 0.05, Tukey's HSD test).

One of the most important properties of hydrogels is their swelling degree, as it directly influences the diffusion of nutrients, cells, and metabolites, as well as the transport and release of drugs [47]. Herein, the degree of swelling was measured by weighing the hydrogel before and after immersion in different media, including phosphate-buffered saline (PBS) at pH 7.4, acetate buffer at pH 4.5, and a solution containing type II collagenase at pH 7.4. In general, the swelling rate did not vary between the PNVCL groups with and without the addition of TC, showing a similar swelling ratio at all evaluation periods in the different solutions evaluated, except for P407 (Fig. 3A–C). P407 exhibited a unique behavior, characterized by a reduction in the swelling ratio over time, which may be attributed to its easy erosion and degradation [48]. Regarding degradation, a slow degradation was observed, with greater material mass loss in acidic media compared to alkaline media (Fig. 3D–F). PNVCL hydrogels have also been associated with pH responsiveness, remaining stable due to the aggregation of the polymer chains under acidic pH conditions and in a swollen and dispersed state at pH 7.4 [49]. However, for both conditions, we observed a slow degradation over time. It can also be observed that the control P407 showed the most significant degradation among the materials tested, which is justified by its easy erosion and degradation, as mentioned above [48]. In contrast, the PNVCL hydrogels, with or without TC, showed a maximum degradation of around 30% within the evaluated period. In addition, the swelling and degradation of hydrogels were evaluated under temperature variations as would occur in the oral environment; however, PNVCL hydrogels appear to be unaffected, as this polymer exhibits excellent stability (Figs. S2 and S3, Supporting Information) [26].

Fig. 3.

Characterizations of hydrogels (n = 5). A) Degree of swelling of hydrogels in PBS (pH 7.4). B) Degree of swelling of hydrogels in acetate buffer (pH 4.5). C) Degree of swelling of hydrogels in collagenase type II (pH 7.4). D) Degradation of hydrogels in PBS (pH 7.4). E) Degradation of hydrogels in acetate buffer (pH 4.5). F) Degradation of hydrogels in collagenase type II (pH 7.4). G) Drug release in PBS (pH 7.4). H) Drug release in acetate buffer (pH 4.5). I) Drug release in collagenase type II (pH 7.4). J, J′) Injectability of hydrogels and images illustrating their flow. K, K′) Lap shear strengths of hydrogels and a schematic illustration of the experimental setup for testing the lap shear strength. Different uppercase letters indicate differences between different groups, and different lowercase letters indicate statistically significant differences between evaluation periods within each group (p < 0.05; Tukey HSD test).

To evaluate drug release, hydrogels were exposed to different media: PBS (pH 7.4), acetate buffer (pH 4.5), and type II collagenase (pH 7.4). The hydrogels were immersed in the respective solutions, and at predetermined time points, the release medium was collected for analysis using a UV-VIS spectrophotometer. Fresh medium was then added for subsequent measurements. Hydrogels without TC incorporation were used as controls. According to the results, drug release was proportional to the concentration of TC loaded into the hydrogels at the different evaluation periods and in the different media (Fig. 3G–I). The release of TC was greater in an enzymatic medium containing type II collagenase, an enzyme associated with inflammation processes. This increased release may be attributed to the enzymatic hydrolysis of PNVCL. It has been shown that PNVCL exhibits higher enzymatic hydrolysis as its molecular weight increases, which may be attributed to the strengthening of inter- and/or intramolecular aggregations [50,51]. Specifically, the molecular weight of the PNVCL synthesized in the present study, with 30.7 10³ g/mol is comparable to the average molecular weight evaluated in the study of Lv et al. (2021) [50], which showed increased enzymatic hydrolysis of the polymer with 32 × 10³ g/mol compared to the polymer synthesized with lower molecular weights such as 1.3 × 10³ g/mol and 1.8 × 10³ g/mol. Additionally, there was a tendency for lower drug release in acidic media, which may also be related to the polymer's greater stability and stronger aggregation in acidic pH conditions [49]. Under temperature variations, no significant differences were observed in TC release (Fig. S4, Supporting Information). However, there is apparent stability and increased release over 14 days, which demonstrates a promising approach for local drug delivery to treat chronic conditions such as peri-implantitis.

For the injectability analysis, P407, PNVCL, and PNCVL with TC 5 mg/mL showed similar results (p > 0.05). In contrast, PNCVL with 25 mg/mL and 50 mg/mL of TC required a greater average force to sustain the movement of the syringe plunger and inject the hydrogel. However, these values are within acceptable limits for clinical application (Fig. 3J–J’). It has been suggested that the maximum acceptable injection force clinically is 40 N. Still, the recommended target is no more than 20 N, with the values found in this study being below the acceptable limit [52]. The adhesion of hydrogels to the substrate or site where they are applied is a crucial property for maintaining their stability and facilitating the release of the loaded drug at the site of the disease. According to the hydrogel adhesion results evaluated by the lap shear test (Fig. 3K–K’), it can be observed that PNVCL, regardless of the TC concentration loading, showed greater shear strength compared to the commercial control P407 (p < 0.05). In addition, although there is no consensus in the literature on the ideal values for the adhesion of hydrogels to dental implant surfaces, the findings are within the range of values reported in the literature for adhesive hydrogels. Li et al. (2020) [53] synthesized a hydrogel with adhesive properties, reporting a shear strength of 4.6 kPa on pig skin. In contrast, Ghosh et al. (2016) [54] described their hydrogel as strongly adhesive to pig skin, with a shear strength of 35 kPa, attributed to the benzaldehyde functional groups. Similarly, Shirzaei Sani et al. (2019) [21] classified their hydrogel as bioadhesive due to a shear strength of approximately 60 kPa. Cheng et al. (2017) [20] also developed a hydrogel coating that exhibited comparable adhesion (∼60 kPa) to titanium substrates. Thus, the newly developed PNVCL-based hydrogel for the controlled release of TC, with an adhesive stress of around 20 kPa, can be considered a biomaterial with the ability to adhere and remain stable on the surface of implants.

2.2. In vitro biological and microbiological analyses

The in vitro biological cytocompatibility level of medical devices and biomaterials must be above 80% (ISO 10993-5:2009) to ensure that proposed new biomaterials are safe for in vivo application. Herein, the metabolic activity of human gingival fibroblasts (HGFs) cells was evaluated when in contact with the 1-day hydrogel eluate using indirect cytotoxicity analysis, as determined by the CCK-8 test. The indirect cytotoxicity analysis of the PNVCL hydrogel (Fig. 4A and B) revealed no cytotoxicity, with superior results compared to P407 (p < 0.05) and comparable to the positive control (p > 0.05). In addition, from the fluorescence microscopy images (Fig. 4C), it can be observed that PNVCL has a positive effect on HGFs, with apparent cell proliferation compared to P407. Despite the morphological changes observed, the significant increase in cell proliferation suggests that the cells remained viable and successfully adapted to the new environment. This behavior indicates a potential bioactive effect of the biomaterial, with no signs of cytotoxicity. Importantly, PNVCL has been shown to present good biocompatibility and desirable affinity with a diverse range of cells [26,55] as well as being non-toxic in a subcutaneous in vivo mouse model [56].

Fig. 4.

In vitro biological analyses (n = 6, 2 independent experiments). A) Cell proliferation examined using CCK-8 (absorbance at 450 nm). B) Cell metabolic activity (%) assessed by CCK-8 after 1 day of indirect contact of cells with 1 day hydrogel eluate, the control group was set as the standard. C) Images obtained from fluorescence microscopy after 1 day of indirect contact of cells with 1-day hydrogel eluate via live/dead cell analysis (green for live cells, red for dead cells). C+ represents the positive control. D) Colony-forming units (log₁₀ CFU/mL) of polymicrobial biofilm developed from saliva inoculum and exposed to the hydrogel with direct application after the supragingival phase. E) Colony-forming units (log₁₀ CFU/mL) of polymicrobial biofilm developed from saliva inoculum and exposed to the hydrogel with direct application after the subgingival phase. F) Colony-forming units (log₁₀ CFU/mL) of polymicrobial biofilm developed from saliva inoculum and exposed to 1-day hydrogel eluate after the supragingival phase. G) Colony-forming units (log₁₀ CFU/mL) of polymicrobial biofilm developed from saliva inoculum and exposed to 1-day hydrogel eluate after the subgingival phase. H) Colony-forming units (log₁₀ CFU/mL) of polymicrobial biofilm developed from saliva inoculum and exposed to 7-day hydrogel eluate after the supragingival phase. I) Colony-forming units (log₁₀ CFU/mL) of polymicrobial biofilm developed from saliva inoculum and exposed to a 7-day hydrogel. J) Colony-forming units (log₁₀ CFU/mL) of polymicrobial biofilm developed from saliva inoculum and exposed to 14-day hydrogel eluate after the supragingival phase. K) Colony-forming units (log₁₀ CFU/mL) of polymicrobial biofilm developed from saliva inoculum and exposed to a 14-day hydrogel. L) Micrographs obtained by SEM of polymicrobial biofilms developed from saliva inoculum and exposed to the 1-day, 7-day, and 14-day hydrogel eluate in the supra and subgingival phases (magnification 2000 × , 15 kV, scale bar = 10 μm). Different capital letters indicate differences between different groups, and bars indicate a significant difference between groups (p < 0.05; Tukey HSD test).

For the in vitro microbiological analyses, commercial substrates (titanium discs) with surface treatment by the HAnano Unitite were used (∼0.20 μm arithmetic roughness; Fig. S5, Supporting Information). For the in vitro analyses, polymicrobial biofilms developed from human saliva as an inoculum were used in both aerobic and anaerobic conditions, as this is a suitable model and clinically relevant tool for mimicking the polymicrobial composition of implant-related infections [57]. Moreover, both conditions used—supragingival and subgingival—represent the clinical onset of the disease, beginning as mucositis at the supragingival level (aerobic), which can progress to peri-implantitis at the subgingival level (anaerobic). After 48 h of aerobic growth, followed by a further 48 h under anaerobic conditions for biofilm formation, the biofilms were scraped with titanium curettes, simulating a preceding step of mechanical biofilm removal before the hydrogel was applied. In addition, the in vitro antimicrobial potential of hydrogels was assessed by direct contact with the hydrogel and indirect tests using the hydrogel eluate. According to the results of the in vitro microbiological tests, PNVCL hydrogel loaded with TC significantly reduced bacterial counts in both direct and indirect experiments under supragingival (aerobic) and subgingival (anaerobic) polymicrobial conditions (Fig. 4D–K). When applied directly (Fig. 4D and E) to biofilms formed under aerobic conditions (supragingival), P407 and PNVCL hydrogels without antibiotic loading showed similar colony-forming units (CFU) compared to the control group (no treatment) (p > 0.05). In contrast, the PNVCL groups loaded with 5 mg/mL and 25 mg/mL of TC showed a CFU reduction of ∼2 log, while PNVCL with 50 mg/mL of TC presented no bacterial growth (p < 0.05). In the subgingival phase, the antibiotic-loaded hydrogel reduced the bacterial count by ∼2 log units, regardless of the concentration of the incorporated drug (p < 0.05). For the 1-day hydrogel eluate (Fig. 4F and G), there were no differences between the control, P407, and PNVCL groups; however, the antibiotic-loaded groups showed no bacterial growth during the supragingival phase. Conversely, in the subgingival phase, the reduction was ∼2 log units for the antibiotic-loaded groups compared to the unloaded ones (p < 0.05), which can be attributed to the higher bacterial levels and the increased complexity of the biofilm structure that developed over a more extended period. For the 7-day hydrogel eluate, in the supragingival and subgingival phases, the PNVCL groups with 5 mg/mL and 25 mg/mL of TC showed ∼2 log reduction. In comparison, the group with 50 mg/mL of TC presented no bacterial growth (p < 0.05) (Fig. 4H and I). The same profile was observed for the 14-day hydrogel eluate (Fig. 4J and K). Qualitatively, SEM images revealed a significant reduction in bacterial clusters over time in all groups treated with TC-loaded hydrogels (Fig. 4L). Although the results were promising, with significant decreases in CFU in both aerobic (supragingival) and anaerobic (subgingival) conditions, the reductions in CFU were smaller in subgingival biofilms, as anaerobic bacteria are less susceptible to TC than aerobic bacteria [58]. Importantly, TC has demonstrated significant efficacy in controlling supragingival biofilms, which are often present in the early stages of peri-implant infections and can help prevent disease progression and microbial shifts. This effect can contribute to the early control of microbial colonization and delay the formation of a dysbiotic and pathogenic biofilm. Moreover, even at lower levels, the reduction at the subgingival level was sustained over 14 days, which can directly result in less tissue damage. In addition, the well-known anti-inflammatory and collagenase-inhibitory properties of TC enhance its therapeutic potential as an adjuvant in the management of peri-implantitis. TC inhibits matrix metalloproteinases (MMPs), key collagenases involved in connective tissue degradation, by chelating divalent cations required for their enzymatic activity [59]. Moreover, in the analyses evaluating the antimicrobial potential of the hydrogel eluate, cumulative release of 1, 7, and 14 days was used in the experiments. In a real clinical condition, where there is no closed system as in in vitro experiments, the eluate in different periods would likely have a different antimicrobial effect. However, these experiments provide valuable results on the maintenance of TC's antimicrobial effect even in the later periods. It is essential to note that these analyses were limited to the number of viable bacteria and did not account for the composition and other characteristics of the biofilm. In this way, a further dynamic and in-depth evaluation of the antimicrobial properties of this new biomaterial was conducted in an in situ human model.

2.3. In situ human microbiological analyses

Given the antimicrobial potential of the PNVCL hydrogel loaded with TC in in vitro assays, as well as its suitable properties for application in the oral environment, an in situ experiment was conducted. For this, healthy volunteers used palatal devices in which titanium discs, with commercial surfaces treated with HAnano Unitite, were exposed to the oral environment to allow biofilm formation and mimic clinical conditions in terms of polymicrobial profile and biofilm biomass (Fig. 5A) [60,61]. The volunteers used the palatal devices for 4 days, and then the discs were removed, scraped with titanium curettes for mechanical debridement, and the hydrogel was applied on the surface of the discs in vitro. According to the results, P407 and PNVCL hydrogels without antibiotic loading showed similar bacterial live cell counts compared to the control group (p > 0.05). PNVCL hydrogel loaded with 5 mg/mL of TC showed a CFU reduction of ∼2 log, while PNVCL with 25 mg/mL and 50 mg/mL of TC showed a reduction higher than 3 log (Fig. 5B) (p < 0.05). In Fig. 5C, the confocal laser microscopy images show the presence of live cells (green), stained with Syto-9, and dead cells (red), stained with a propidium iodide solution, illustrating bacterial death in the groups treated with TC-loaded hydrogel at the different concentrations evaluated. The in-situ biofilms were also susceptible to the antimicrobial and bacteriostatic action of TC. Therefore, the in vitro microbiological results were confirmed in the in situ analyses, demonstrating that the proposed antimicrobial hydrogel also has great potential, even in dynamic situations more similar to those that occur clinically in the oral environment.

Fig. 5.

In situ microbiological analyses with biofilm formed on the surface of HAnano Unitite discs to test the antimicrobial effect of the hydrogel (n = 6). A) Representative scheme of the intraoral devices. B) Colony-forming units (log₁₀ CFU/mL) of biofilms formed in situ and exposed to the hydrogel. C) Representative confocal laser microscopy images. Living cells (green) were stained with Syto-9, and dead cells (red) were stained with propidium iodide solution. D) Beta diversity by principal coordinate analysis (PCoA) in each group. E) Relative abundance of the top dominant genus. F) Shannon diversity in each group. G) Relative abundance of microbial species present only in the control groups (P407 or PNVCL). H) Microbial genus increased fivefold or more in the control groups (P407 or PNVCL). I) Relative abundance of the genus related to peri-implant infections in each group. Statistically significant differences between groups are indicated by bars (∗∗∗∗p < 0.0001; Tukey HSD test).

To further evaluate the effect of the PNVCL hydrogel loaded with TC on biofilms formed in situ, 16S RNA profiling was carried out. After sequencing and initial bioinformatic analysis, a total of 336 amplicon sequence variants (ASVs) were taxonomically assigned to the bacterial species level across all groups. To explore the patterns of similarity and dissimilarity among the samples from the groups, principal coordinate analysis (PCoA) was used. Although there is some discrepancy among the samples, most were grouped, showing high similarity (Fig. 5D). This outcome may be attributed to the high interindividual variability, as each subject harbors a unique microbial profile that differs from that of others.

Regarding the nine most abundant microbial genera, most samples showed Streptococcus and Neisseria as the dominant taxa (Fig. 5E). Interestingly, in the groups treated with TC, there was an apparent reduction in the proportion of Veillonella species compared to the control groups. In terms of richness, alpha diversity, as assessed by the Shannon Index, showed no significant differences among the groups (Fig. 5F), indicating that the number of species present in each group was similar. Although no significant differences were observed in Shannon diversity among the experimental groups (Fig. 5F), this finding is consistent with the nature of in situ models, where the individual's own oral environment influences the microbiota. The high interpersonal variability and the preservation of ecological conditions in the in situ setting may have contributed to the maintenance of overall microbial diversity, even in the presence of antimicrobial interventions. Since there was no difference between 25 and 50 mg/mL in live bacterial cell counts (Fig. 5B), we investigated the effect of 25 mg/mL on bacterial profile modulation compared to the controls (P407 and PNVCL). Significantly, some bacterial species were suppressed in the PNVCL–25 mg/mL group but were present in the control groups (Fig. 5G), including peri-implant disease-related species such as Fusobacterium spp., Prevotella spp., and Staphylococcus spp. Clinical evidence has demonstrated the presence and role of Fusobacterium species in peri-implant mucositis and peri-implantitis [62], identifying them as important pathogens associated with the colonization of other key oral microorganisms related to disease. Moreover, both genera, Prevotella spp. and Fusobacterium spp., have been found at high levels in peri-implantitis sites prior to treatment [7]. These findings highlight the role of PNVCL loaded with TC in not only reducing bacterial levels but also modulating the microbial profile, decreasing the abundance of important peri-implant pathogens. As extensively discussed in the literature, biofilm composition plays a key role in disease progression and tissue damage; therefore, antimicrobial strategies should aim not only to reduce bacterial levels but also to modulate the biofilm profile toward a health-associated composition. In this sense, 13 bacterial genera were found to be increased at least fivefold in one or both control groups (P407 or PNVCL) compared to PNVCL–25 mg/mL (Fig. 5H). Among them, Tannerella spp. was also reduced. In fact, when evaluating bacterial genera highly associated with peri-implant diseases, the control groups showed samples with higher levels of Tannerella spp. and Aggregatibacter spp. compared to the TC group (Fig. 5I). Tannerella forsythia has been recognized as an important biomarker of peri-implantitis and is strongly associated with peri-implant tissue damage [62]. Moreover, Aggregatibacter spp. has been described as an important oral pathogen capable of inducing an exacerbated host response and tissue damage [63]. Therefore, in biofilms grown in the oral cavity of volunteers, which mimic the polymicrobial profile, PNVCL–25 mg/mL was able to slightly modulate biofilm composition, reducing the abundance of putative oral pathogens known to induce peri-implant disease progression and tissue destruction, highlighting its ability to control microbial growth at a species-specific level.

2.4. In vivo analyses

To confirm the findings of the antimicrobial hydrogel loaded with TC in an in vivo condition, with greater complexity and encompassing a dynamic environment, a rat tibia model was used (Fig. 6A). For this, implants treated with HAnano Unitite, the same used in the in vitro and in situ assays, were contaminated (cervical threads) under in vitro conditions by anaerobic polymicrobial biofilms developed from saliva, as in the in vitro study [57]. After the biofilm was formed, the cervical threads of the implants were scraped using a titanium curette. This step was used to reproduce the mandatory clinical step of mechanical removal of biofilm from the surface of dental implants, in accordance with the procedures performed in the in vitro and in situ studies [64,65]. After this step, implants were immediately implanted in the rats' tibia, followed by a bone defect creation around the contaminated cervical threads in an attempt to mimic the bone loss present in peri-implantitis (Fig. S6, Supporting Information). Then, hydrogels were applied in the bone defect area, and the surgical wound was sutured. In these in vivo experiments, implants with contamination in the cervical threads were used. The groups consisted of control contaminated implants without any treatment, contaminated implants with the application of the hydrogel without TC loading, and contaminated implants with the application of the hydrogel containing 25 mg/mL of TC. This concentration was selected because it yielded more promising results in the antimicrobial analysis of the in situ study, and it also represented the intermediate concentration tested. Additionally, selecting a single TC concentration helped reduce the number of animals used, in accordance with ethical guidelines.

Fig. 6.

In vivo microbiological analyses (n = 6). A) Representative scheme of the in vitro contamination of implants and subsequent installation in the tibia of rats with the creation of a bone defect to simulate bone loss as occurs in peri-implantitis and the application of hydrogel. B) Colony-forming units (log₁₀ CFU/mL) of polymicrobial biofilm developed from saliva inoculum and installed on the tibia of rats after the subgingival phase with subsequent exposure to hydrogel. C) Shannon diversity in each group. D) Relative abundance of species present only in the control groups. E) Differential heat trees comparing ASV abundance values between the groups. Color indicates the dominant species in each substrate, Control/PNVCL (red) or PNVCL-25 mg/mL (blue), on a log₂ median proportion scale. Node size indicates normalized ASV counts. F) Expression of IL-4, IL-1β, IL-6, IFN-γ, IL-17 and TFN-α cytokines. Bars indicate statistical differences between groups (∗p < 0.05; ∗∗p < 0.01; Tukey HSD test).

After 28 days, the animals were euthanized, and the results of PNVCL hydrogel loaded with TC in a dynamic in vivo environment were assessed. It can be observed that the microbiological analysis results, as indicated by CFU counts, align with the in vitro and in situ findings. The PNVCL hydrogel loaded with TC promoted a significant reduction in the count of viable bacteria compared with control and PNVCL without TC loading (p < 0.05) (Fig. 6B). 16S RNA sequencing analysis was also performed in the in vivo study, and a total of 64 ASVs were taxonomically assigned to the bacterial species level across all groups. For microbial richness, alpha diversity, as assessed by the Shannon Index, showed no significant differences among the groups (Fig. 6C). Eight bacterial species were found only in the control groups compared to the TC group (Fig. 6D), including Staphylococcus spp., which has been extensively studied for its role in implant device infections [66]. In fact, the differential tree showed group-specific differences in species abundance, suggesting a modest effect of TC in modulating the microbial profile (Fig. 6E). It is essential to note that the late period of analysis (28 days) and the presence of a distinct microbiota in rats compared to the human microbiota are factors to consider when interpreting these data. Most of the relative percentages of the genus were unclassified bacteria, which also makes it difficult to interpret this analysis. Animal models are much more amenable to extensive investigations than patients and therefore provide an opportunity not only to reproduce the main features of a disease but also to provide strong evidence, especially for clinical purposes. However, there are intrinsic limitations to the tibia model used, including a region that is not exposed to the oral environment, such as saliva and food. Furthermore, the tibia is also an environment that does not allow microbial recolonization, unlike the oral cavity, and is continuously exposed to external factors. In the present study, in vivo drug release was not evaluated, which is recognized as a limitation since in vitro experiments may not represent the dynamic in vivo conditions. Despite these limitations, this tibia model enables us to assume that the PNVCL hydrogel loaded with TC retains its antimicrobial potential, resulting in a reduction in viable bacteria, even in an environment with greater complexity and dynamics compared to in vitro and in situ studies, as well as in an environment with a different microbial composition.

The expression of pro-inflammatory and anti-inflammatory cytokines was also assessed in the in vivo study. It can be observed that for most of the cytokines evaluated, there were no statistically significant differences between the studied groups (Fig. 6E). Although not statistically significant, IL-1β tended to be higher in the control group. This tendency may be explained by the fact that this pro-inflammatory cytokine plays a key role in host defense against infection, and the control group was infected but did not receive any treatment [67]. In contrast, although not statistically significant, IL-6 and IL-17A tended to be higher in PNVCL and TC-loaded PNVCL than in the control group. These cytokines are also pro-inflammatory, and their role in the foreign body response is not well understood; however, this may provide a hypothesis for the tendency of higher secretion of these cytokines [68]. These results must be analyzed with caution, as the late period of analysis (28 days) after application of the hydrogel may have been an influencing factor. Indeed, the initial inflammatory response in earlier periods would have provided more evidence of the inflammatory response of the hydrogels, as well as the material's potential to reduce implant contamination and local inflammation.

Bone neoformation was evaluated in the in vivo model through microcomputed tomography (micro-CT) and histology analyses (Fig. 7). It was observed that the hydrogel increased bone formation compared to the control group (Fig. 7A). One hypothesis to explain the promising results in bone neoformation is that the hydrogel has served as a scaffold for cells, enhancing their adhesion and proliferation. Furthermore, the porous and interconnected three-dimensional structure of the biomaterial may also have facilitated the diffusion of nutrients to these cells, aiding in the process of bone neoformation [69]. Additionally, the influence of TC on calcium metabolism, acting as a modulator of osteoclast generation and bone mineralization, can be attributed to the superior performance of PNVCL hydrogel loaded with TC [70]. The histological results corroborated those of the micro-CT analysis (Fig. 7B), showing excellent results for new bone formation in the PNVCL hydrogel compared with the control group in both the bone defect area and the bone-implant interface. The combined use of the hydrogel as a platform for cells, associated with TC, which offers antimicrobial potential and influences calcium metabolism and chelation, has produced positive outcomes not only in the area of the bone defect, but also at the bone-implant interface. The hydrogel promoted significant improvements in bone volume, bone volume percentage, trabecular thickness, and the newly formed bone area in the region of the bone defect around the implant's cervical threads (Fig. 7C–F), highlighting its regenerative potential with osteogenic properties. These effects may be attributed to the biomaterial's ability to stimulate cell proliferation and new bone formation in response to its composition. At the bone-implant interface, significant improvements were observed in bone-implant contact and newly formed bone within the defect (p < 0.05), with no differences in bone volume or trabecular thickness (Fig. 7G–J). Altogether, these results indicate the beneficial effects of the hydrogel, particularly in enhancing bone regeneration in peri-implant defects and improving bone-implant integration. In compromised implant sites, this biomaterial suggests potential for stabilizing failing implants by promoting bone fill in critical regions, such as the cervical threads, which are essential for long-term implant retention and functional recovery.

Fig. 7.

In vivo bone formation analyses (n = 6). A) Representative micro-CT images. B) Photomicrograph of control, PNVCL, and PNVCL/TC 25 mg/mL groups with hematoxylin and eosin staining for histological analysis (WB: woven bone; TB: trabecular bone; CT: connective tissue). C) Mean values and standard deviation of bone volume (mm³) in the bone defect area. D) Mean values and standard deviation of percent bone volume (%) in the bone defect area. E) Mean values and standard deviation of trabecular thickness (mm) in the bone defect area. F) Mean values and standard deviation of newly formed bone (μm³) in the bone defect area. G) Mean values and standard deviation of bone volume (mm³) in the bone-implant interface. H) Mean values and standard deviation of trabecular thickness (mm) in the bone-implant interface. I) Mean values and standard deviation of bone-implant contact (μm) in the bone-implant interface. J) Mean values and standard deviation of newly formed bone (μm³) in the bone-implant interface. Bars indicate a significant difference between the groups (∗p < 0.05; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001; Tukey HSD test).

In addition, immunohistochemical analysis revealed distinct staining patterns for osteopontin (OPN) and osteocalcin (OCN) among the experimental groups (Fig. 8). As a matrix-associated protein, OPN immunoreactivity was mainly observed at the reversal line separating the residual bone from the newly formed woven bone [71]. OPN immunoreactivity was markedly higher in the control group, whereas a reduction in OPN labeling was observed in the PNVCL group. Notably, specimens treated with PNVCL hydrogel loaded with TC exhibited minimal to absent OPN immunostaining (P < 0.05). OCN, the major non-collagenous protein of the bone matrix and a marker of mineralization [[72], [73], [74]], was predominantly detected in mature osteoblasts and in osteocytes entrapped within the extracellular matrix. OCN immunostaining was scarcely detected in control specimens, exhibiting predominantly absent or low labeling. The PNVCL group demonstrated an increase in OCN immunoreactivity, characterized by a moderate staining pattern. The highest OCN expression was observed in the PNVCL loaded with TC group, with a predominance of intense immunostaining (P < 0.05). This analysis was essential for correlating immunohistochemical findings with histomorphometric and microtomographic data, which demonstrated that the test group (PNVCL/TC 25 mg/mL) promoted greater bone tissue neoformation and superior bone quality in terms of maturation. These findings were supported by lower OPN expression and higher OCN expression in the PNVCL/TC 25 mg/mL group compared with the other experimental groups. Collectively, the results indicate that, beyond its antimicrobial activity, restoring a previously contaminated microenvironment, the hydrogel promotes the recruitment and activation of cells from the osteoblastic lineage, thereby enhancing bone regeneration.

Fig. 8.

In vivo immunohistochemical analysis (n = 6). (A) Representative immunolabeling images of the Control, PNVCL, and PNVCL/TC 25 mg/mL groups showing osteopontin (OPN) and osteocalcin (OCN) expression (immunopositive cells indicated by black arrows). (B) Mean scores (0–3) ± standard deviation for OPN immunostaining. (C) Mean scores (0–3) ± standard deviation for OCN immunostaining. Bars indicate statistically significant differences between groups (p < 0.05; Tukey HSD test).

Our results show that the high swelling of the PNVCL hydrogel allowed the loading of TC into its three-dimensional and porous structure, which may also have facilitated the diffusion of cells and nutrients, thereby enhancing its biocompatibility and osteogenic potential. Furthermore, the controlled release of TC for an extended period, associated with a slow degradation and good adhesiveness, may have supported antimicrobial activity and new bone formation. The hydrogel may have acted as a cellular anchoring point, and as it degraded over time, new bone tissue formed in the defect area. However, future studies evaluating the in vivo degradation of the PNVCL hydrogel, as well as in vivo models that more closely mimic the clinical conditions of peri-implantitis, are essential to provide sufficient evidence for the clinical application of this biomaterial. After all, the oral environment, with its salivary flow and constant temperature changes, could affect the stability of PNVCL hydrogels, as it involves reversible self-assembly when temperatures fluctuate below or above its LCST. In the case of its application for treating peri-implant mucositis, the absence of bone loss and the lack of pockets and sites for anchoring the hydrogel could facilitate its detachment, as the hydrogel would be more exposed to temperature changes in the oral cavity, and greater reversibility events could occur, affecting its adhesiveness. However, the antimicrobial potential of the PNVCL hydrogel loaded with TC could be beneficial in reducing the inflammation of peri-implant tissues in mucositis, as well as providing a treatment that can be reapplied when necessary.

In contrast, in cases of peri-implantitis with bone loss, more available sites for adhesion and anchorage of the hydrogel would be present, and the material could also facilitate treatment by aiding bone neoformation with simultaneous antimicrobial activity. Furthermore, this biomaterial could be applied in both non-surgical and surgical approaches for the treatment of peri-implant infections. Despite the limitations of this study, including the use of an animal model that does not fully replicate clinical conditions, further analysis is needed to deepen our understanding of the behavior of this biomaterial. In general, the proposed hydrogel showed highly effective antibacterial activity in vitro, in situ, and in vivo, as well as osteogenic properties for the control and treatment of peri-implant infections.

3. Conclusion

We herein designed a thermoresponsive hydrogel loaded with tetracycline (TC) for the control and treatment of peri-implant infections. This platform was strategically designed to address a critical clinical need, underscoring its potential for future translational and clinical applications. The poly(N-vinylcaprolactam) hydrogel displayed promising results, as it can remain in the diseased area for an extended period due to its adhesiveness, allowing for the controlled release of TC and reducing the viability of bacteria. Notably, the newly developed hydrogel was not only able to reduce microbial levels in all tested models but also modulated microbial composition, reducing some important oral pathogen species. Additionally, the significant formation of new bone with a higher degree of maturation observed in bone defect areas and at the bone-implant interface suggests that this biomaterial is a valuable treatment option, particularly for compromised implant sites that require enhanced healing and bone regeneration. With its combined antimicrobial and osteogenic properties, this novel hydrogel formulation offers a fast, safe, and highly effective treatment for peri-implant infections.

4. Experimental section

4.1. Synthesis of the hydrogel

The PNVCL hydrogel was synthesized according to a previously defined methodology with some modifications [45]. Initially, 2 g of N-vinylcaprolactam monomer (±98%; Sigma-Aldrich) was added to 40 mL of distilled water. Then, after the solution reached 70 °C, 500 μL of the initiator 2,2′-azobis (2-methylpropionitrile) (AIBN, 98%; Sigma-Aldrich) was added. The reaction was carried out at 70 °C under a nitrogen atmosphere for 4 h to facilitate polymerization. The PNVCL was purified against distilled water, lyophilized, and stored at 4 °C before use [26]. The hydrogel without the addition of TC was used as a control, and a commercial thermosensitive hydrogel, Poloxamer 407 (26% w/w) (Sigma-Aldrich), was used as a commercial control [48]. For the experimental groups, TC was added to the hydrogel at three different concentrations: concentration (1) at the minimum bacterial concentration (MBC), concentration (2) 5 times the MBC, and concentration (3) 10 times the MBC. PNVCL hydrogels were prepared at a concentration of 20%. Given the instability of TC in the air and its sensitivity to light, the hydrogels were prepared immediately before each experiment [35].

4.2. Minimum bactericidal concentration of tetracycline

Saliva samples were stimulated with Parafilm (Parafilm M, American Can Co., Neenah, WI, USA) and collected from healthy volunteers. The saliva samples were incubated in universal fluid medium (FUM) supplemented with 67 mmol/L Sorensen's buffer, pH 7.2 ("modified universal fluid medium", FUMm) and 10% sterile brain heart infusion (BHI) broth (Difco Laboratories) in a 1:10 ratio for 12 h at 37 °C under anaerobic conditions. The inoculum was then adjusted to an optical density (OD) of 0.1 at 550 nm using a spectrophotometer (Spectronic 20, Bausch & Lomb, Rochester, NY, USA). After adjusting the OD, 100 μL of the inoculum was added to different concentrations of tetracycline hydrochloride (Sigma-Aldrich, St. Louis, MO, USA) and incubated in 96-well plates at 37 °C for 24 h under anaerobic conditions. The minimum bactericidal concentration (MBC) was determined by seeding 20 μL of the solution from each well into Columbia Blood Agar (CBA) culture medium supplemented with 5% defibrinated sheep's blood. The plates were incubated at 37 °C for 72 h, and the MBC was defined as the lowest concentration capable of causing bacterial death, with no bacterial growth after incubating the plates for 3 days [75]. From the MBC results, the TC concentration was defined to be incorporated into the hydrogel, in addition to 5 and 10 times the MBC concentration, which were also evaluated.

4.3. Scanning electron microscopy of the hydrogel

The morphological aspects of the networks and the polymeric matrices of the PNVCL hydrogels were observed using SEM. The morphology of cross-sections of hydrogels at a voltage of 2.02 kV was examined at a magnification of 200 × using a ZEISS Sigma 300 microscope (Carl Zeiss Microscopy GmbH).

4.4. Fourier transform infrared spectroscopy

FTIR was carried out to obtain the spectra and analyze the expected functionalities. The attenuated total reflection (ATR) method was employed using a TENSOR 27 (Bruker, Germany) spectrometer (4000-500 cm⁻¹), a HeNe laser source, and a DLaTGS detector with a resolution of 4 cm⁻¹ [76,77].

4.5. Nuclear magnetic resonance spectroscopy

Nuclear magnetic resonance spectroscopy (¹H NMR) was carried out on a Bruker Fourier 600 MHz spectrometer measuring samples dissolved in deuterium chloroform [78].

4.6. Differential scanning calorimetry

DSC studies were performed with about 10 mg of lyophilized samples (TA Instrument DSC 2920 Modulated DSC). Aluminium pans were used to contain the samples for the DSC experiment. All samples were examined under a pure nitrogen atmosphere. During testing, all samples were ramped from 50 °C to 250 °C at a rate of 10 °C/min. The analysis was conducted on all samples, and the glass transition temperature (Tg) was recorded [79].

4.7. Critical solution temperature, gelation time, and rheological properties

To analyze the critical solution temperature at which the polymer undergoes phase separation, an AR2000 Advanced Rheometer (TA Instruments, Asse, Belgium) with a steel Peltier platform (D = 20 mm) was used. The sol-gel transition temperature (Tsol-gel) was determined as the point at which the storage modulus (G′) and loss modulus (G″) intersect during the temperature sweep measurements. The G′ and G″ values of each group at 37 °C were recorded after 2 min of steel Peltier platform temperature stabilization [48]. To test the sol-gel transition temperature and mechanical properties of the hydrogels, a time sweep program was used, with a heating rate of 2 °C/min, starting at 4 °C and increasing to 37 °C, at a loading frequency of 1 Hz and a voltage of 2% [48].

4.8. Size exclusion chromatography

The molecular mass distribution of the polymer was determined using an OMNISEC size exclusion chromatograph associated with Viscotek RImax refractive index detectors (Malvern Instruments). The samples (5.0 mg) were diluted in THF and filtered through 0.45 μm PTFE membranes. Three Shodex chromatographic columns, model KF806M (Showa Denko, Tokyo, Japan), were used, with the temperature maintained at 40 °C. A THF injection flow rate of 1.0 mg mL⁻¹ was used. The calibration curve was obtained using polystyrene standards with molecular masses ranging from 4.2 × 10⁶ to 1200 g mol⁻¹ (Varian, Palo Alto, CA) [80].

4.9. Swelling and degradation ratio

To assess the swelling ratio, 200 μL of hydrogel was prepared in Eppendorf tubes. The weight of the hydrogel was measured to obtain the initial weight (Wi). Then, the samples were immersed in 1 mL of PBS solution (pH 7.4), acetate buffer (pH 4.5), or a solution containing type II collagenase at a concentration of 20 μg/mL and incubated at 37 °C. At predetermined time intervals (1, 3, 7, and 14 days), the solutions were removed and the wet hydrogel samples were weighed (Wf) [81]. The percentage swelling ratio (%) of the hydrogels was calculated using the following equation: Swelling ratio (%) = (Wi-Wf)/Wi × 100. For degradation analysis, after each evaluation time of the swelling analysis, the samples were frozen at −80 °C and then lyophilized for 4 days. Lyophilized samples were weighed to obtain the final weight (Wf). The degradation rate was determined using the equation: Degradation rate (%) = (Wi-Wf)/Wi × 100 [21,81,82]. Considering the reversibility of the sol-gel state of the PNVCL hydrogel, additional analyses of swelling and degradation ratio were also carried out with temperature variation to mimic the temperature variations that occur in the oral cavity. For this, the samples were incubated every 24 h for 1 h at 4 °C, and for the remainder of the experiment, they were incubated at 37 °C under constant stirring at 90 rpm.

4.10. Tetracycline release

A UV-VIS spectrophotometer was used to determine the release of TC. For this, hydrogel samples were immersed in 1 mL of PBS (pH 7.4), acetate buffer (pH 4.5), or a solution containing type II collagenase to simulate different conditions (health and inflammation) and kept under agitation at 90 rpm at 37 °C for 1, 3, 7, and 14 days. At each evaluation time, 1 mL of the solution was collected, and an additional 1 mL of solution was added for the subsequent evaluation periods. A standard curve of TC was prepared to determine the TC concentrations released from the hydrogels, as well as the method's detection limits [35]. Considering the reversibility of the sol-gel state of the PNVCL hydrogel, additional analyses were also conducted with temperature variation, as previously described.

4.11. Hydrogel injectability

The injectability test was carried out on a material testing machine (Lloyd Instruments; Ametek, Largo, USA) using a compression mode. 1 mL of hydrogel solution was loaded into a plastic syringe (Becton Dickinson, UK) equipped with a 22G needle, and the syringe was positioned on the dynamometer stand with its tip downwards. The plunger of the syringe was placed in contact with a load cell (rated at 100 N). The test was carried out at a speed of 10 mm/min, which is representative of the manual delivery of materials with a syringe to patients. The average force required to sustain the movement of the plunger and expel the syringe's contents was measured (N). Milli-Q water was used as a control [48].

4.12. Hydrogel adhesion

The adhesion of the hydrogel to a commercial substrate (HAnano Unitite surface donated by the company S.I.N. Implants) was assessed using a shear test with modifications [20,21]. For this, two discs were positioned in parallel with a standardized amount of hydrogel between them (200 μL). Evaluation was then carried out using an Instron 5542 equipment equipped with a 100 N load cell at a speed of 1.3 mm/min to obtain shear strength data (in kPa).

4.13. In vitro cell culture test - indirect cytotoxicity

Human gingival fibroblasts were isolated and cultured from the oral mucosa of a periodontally healthy donor, in accordance with an approved ethical protocol (CMO Radboudumc; dossier #2017–3252). Dulbecco's modified Eagle's medium (DMEM, Sigma Chemical Co.) supplemented with 10% fetal bovine serum and 100 IU/mL penicillin was used for cell growth in 75 cm² cell culture flasks at 37 °C and 5% CO₂. After the cells reached 90% confluence, they were washed in PBS, released using Accutase solution (Sigma-Aldrich), and resuspended in DMEM culture medium. To assess the indirect cytotoxicity of the hydrogel, we followed the recommendations of ISO 10993. Initially, cells were seeded at a concentration of 1 × 10⁵ cells/well and allowed to adhere and grow during incubation at 37 °C for 24 h. After an incubation period, cells were washed in PBS, and a new culture medium containing 100 μl of the 1-day hydrogel eluate was added to each well. Then, cells were incubated with the hydrogel eluate at 37 °C for 24 h. Next, the Cell Counting Kit-8 (CCK-8) assay (Abcam plc, China) was used according to the manufacturer's protocol. Cells without exposure to the hydrogel eluate served as the positive control and were used for data calculation. Cell metabolic activity (%) was expressed using the following formula: $\%ofmetabolicactivity=\frac{100\times ODexperimental}{ODcontrol}$. Cell viability was assessed using the LIVE/DEAD Cell Imaging Kit (Invitrogen, Life Technologies) according to the manufacturer's instructions. The evaluation was conducted after indirect contact of cells with a 1-day hydrogel eluate. Images were acquired using a fluorescence microscope (Zeiss AxioImager Z.1, Carl Zeiss Microscopy GmbH). For positive growth control, cells were incubated in culture medium. As a negative control, cells were incubated in culture medium supplemented with 9% Triton to induce cell death [83].

4.14. In vitro microbiological assay

The study was approved by the local Research and Ethics Committee (C.A.A.E. 49523421.0.0000.5418). For the microbiological experiments, sterile discs (Ø6.0 × 3.0 mm) from HAnano Unitite were used. Discs were transferred to 24-well culture plates and modified universal fluid medium (mFUM) [84], supplemented with 10% brain heart infusion (BHI) (Becton-Dickinson), 10% sucrose, and 10% bacterial inoculum of unfiltered saliva [57].Biofilms were grown under aerobic (supragingival) and anaerobic (subgingival) conditions. Stimulated human saliva (unfiltered) in the supragingival phase was used as the bacterial inoculum (OD = 0.1 at 550 nm) [57]. Initially, the biofilms were cultivated under aerobic conditions (10% CO₂, 37 °C) for 48 h, simulating a supragingival condition. Constant exposure to sucrose was used to form biofilms enriched with exopolysaccharides [61,85]. Every 24 h, the discs were washed in 0.9% NaCl, and new culture medium supplemented with 10% BHI and 10% sucrose was added. After 48 h of biofilm growth in phase 1, the biofilms were transferred to anaerobic conditions (10% H₂, 5% CO₂, 85% N₂, 37 °C), simulating a subgingival condition (phase 2). For this, the biofilms were exposed again to a new bacterial inoculum using human saliva as a source of anaerobic pathogens (OD = 0.1 at 550 nm), and the biofilms were allowed to develop for an additional 48 h in mFUM media. Every 24 h, the discs were washed in 0.9% NaCl, and fresh culture medium was added. After phases 1 (supragingival) and 2 (subgingival), biofilms were scraped with titanium curettes, simulating a preceding clinical step of mechanical biofilm removal [64,65]. Afterwards, discs were washed in 0.9% NaCl, and 200 μL of hydrogel was applied directly onto the disc surfaces. Culture plates were incubated at 37 °C to facilitate gelation of the hydrogel, and fresh culture medium was then added. The samples were incubated for 24 h in contact with the hydrogels, and subsequent analyses were then carried out. Additionally, the indirect antimicrobial activity of the hydrogels was evaluated by exposing samples to the hydrogel eluate for 1, 7, and 14 days. For this, after the discs were scraped with titanium curettes and washed in 0.9% NaCl, samples were incubated with the hydrogel eluate for 24 h. Finally, after incubation periods, the direct and indirect antimicrobial activity of the hydrogels was evaluated by colony-forming units (CFU) count and scanning electron microscopy.

4.15. Colony-forming unit analysis

After incubation with the hydrogel, the discs were transferred to cryogenic tubes containing 1 mL of 0.9% NaCl, vortexed (AP-56, Phoenix) for 10 s, and then sonicated (7 W for 30 s) using a S 150D sonicator (Branson Ultrasonics Corp.). An aliquot of 100 μL of the sonicated suspension was sequentially diluted in 900 μL of 0.9% NaCl, and 10 μL of each dilution was seeded in duplicate on Columbia Blood Agar (CBA) medium supplemented with 5% defibrinated sheep's blood. The CBA plates were incubated under aerobiosis (10% CO₂ at 37 °C for 48 h) for samples from supragingival (aerobic phase), while samples from subgingival (anaerobic phase) were incubated under anaerobiosis (10% H₂, 5% CO₂, 85% N₂, 37 °C). CFU counts were obtained using a stereoscopic microscope [86].

4.16. Scanning electron microscopy

Due to the high-water content in the hydrogel composition and the difficulty in visualizing bacteria in samples with direct application of the hydrogel, only samples treated with the hydrogel eluate were evaluated using SEM. After incubation with hydrogel eluate, biofilms were fixed in Karnovsky solution (2% formaldehyde, 2.5% glutaraldehyde, 0.1 M sodium phosphate buffer solution; pH 7.2) for 2 h, dehydrated in baths with increasing concentrations of ethyl alcohol (50%, 60%, 70%, 90%, and three baths of 100%) for 10 min each, and dried at room temperature. Then, samples were mounted on stubs, sputter-coated with gold, and examined under a scanning electron microscope (JEOL JSM-5600LV) at an accelerating voltage of 15 kV [86].

4.17. In situ microbiological study

Upper arch models of the adult volunteers were obtained for the preparation of the palatal devices. Exclusion criteria included smokers, individuals with salivary flow below normal (stimulated <0.7 mL/min), those who had used antibiotics within the two months before the study, and volunteers with periodontal problems. The devices were fabricated in acrylic resin and contained six samples per device. Before using the palatal device, each volunteer received a list of instructions and bottles containing a 20% sucrose solution. The volunteers were instructed to remove the device from the oral cavity and add one drop of the 20% sucrose solution to each disc, four times a day (at 8:00 a.m., 11:00 a.m., 3:00 p.m., and 7:30 p.m.). After dripping the sucrose, the devices had to be left to rest for 5 min before being returned to the oral cavity, for sucrose diffusion through the biofilm. During the palatal device use, the volunteers were instructed to perform oral hygiene three times a day with fluoride toothpaste and to use the device continuously, except when eating or drinking. The volunteers were asked to clean the device without brushing the area around the discs to prevent removing the biofilm that had formed. During this period, the device was kept in its plastic case, wrapped in moistened gauze.

The volunteers were also instructed not to use mouthwash. No restrictions were assigned on the volunteers' diet [87,88]. On the morning of the fourth day of palatal device use, the discs containing biofilm were removed and washed in 0.9% NaCl. Before applying the hydrogel, the disc surfaces were scraped using a titanium curette. The discs were rewashed in 0.9% NaCl and transferred to 24-well polystyrene culture plates, where they were exposed to 200 μL of hydrogels under in vitro conditions. After hydrogel gelation, the new culture medium was added, and the culture plates were incubated for 24 h at 37 °C. Subsequently, biofilm analyses were performed, including CFU count, sequencing analysis, and confocal laser scanning electron microscopy.

4.18. Analysis of biofilm colony-forming units and composition by 16S rRNA gene sequencing

After 24 h of biofilm incubation with the hydrogel, samples were evaluated for CFU, as described above. Evaluation of the microbiome in all samples collected after the application of the hydrogel was carried out using 16S rRNA sequencing technology. Genomic DNA was extracted from pellets using the ZymoBIOMICS-96 MagBead DNA Kit (Zymo Research, Irvine, CA) on an automated platform. DNA quantity and quality were assessed by absorbance (A260/A280) using a NanoDrop One/OneC (Thermo Fisher), quantified and amplified by RT-PCR OPUS (Bio-Rad), fragment quality evaluated during library preparation with the 4200 TapeStation (Agilent), and pooled libraries quantified with a Qubit 4 Fluorometer (Thermo Fisher). Bacterial 16S rRNA gene sequencing was performed using the Quick-16S NGS Library Prep Kit (Zymo Research), targeting the V3–V4 region of the 16S rRNA gene. The final pooled libraries were cleaned using the Select-a-Size DNA Clean & Concentrator (Zymo Research) and quantified using TapeStation and Qubit. Libraries were sequenced on an Illumina MiSeq using a v3 reagent kit (600 cycles). Amplicon sequence variants (ASVs) were inferred from raw reads using DADA2. ASV counts were normalized with Metacoder, and taxonomic assignment was performed using a custom DECIPHER classifier based on the Human Oral Microbiome Database (eHOMD). Alpha and beta diversity analyses were conducted with Metacoder [89]. Comparisons were made based on relative abundance, Alpha diversity, Beta diversity, and identification of dominant genera and species.

4.19. Biofilm analysis by CLSM

The three-dimensional morphology and presence of live and dead cells were analyzed by CLSM (Zeiss LSM 800, Jena, Germany) operating system (ZEISS ZEN 2.6 system). Cells were stained with the Live/Dead cell viability kit (Thermo Scientific, Waltham, MA, USA, L7012). Live cells were stained green with SYTO-9 reagent (480-500 nm), and dead cells were stained red with propidium iodide (490-635 nm). Biofilms were incubated in a dark environment for 20 min at room temperature, and then images were captured in CLSM [57].

4.20. In vivo analyses

The study was in accordance with the Ethical Principles of Animal Experimentation (Protocol FOA/UNESP 265-2023), and followed the ARRIVE guidelines 2.0 for animal studies [90]. For the in vivo analyses, a rat tibia model was used. Male Wistar rats (approximately 350 g) were housed in cages under controlled conditions (22 ± 2 °C, 12 h light/dark cycle). They received solid food with free access to water, except during the 12 h preceding surgery, when they were fasted for food only. For the experiments, all animals were randomly assigned to groups by lottery and distributed by a third party to ensure unbiased allocation. Initially, the implants (∅1.22 × 2.7 mm) treated with HAnano Unitite, the same surface used in the in vitro and in situ studies, were contaminated (cervical threads) under in vitro conditions using an anaerobic polymicrobial biofilm developed from saliva, as in the subgingival phase of the in vitro study described above [57]. For this, the implants were placed in 24-well polystyrene plates using a device made from orthodontic thread to keep them suspended, and only the cervical third of the implant was immersed in the culture medium to prevent contamination. The biofilm was developed following all the steps previously described in the in vitro microbiological experiment. After the biofilm was formed, the implants were scraped using a titanium curette as a mandatory clinical step of mechanical removal of biofilm. Then, the implants were washed in 0.9% NaCl and immediately installed in the animal's tibia.

4.21. Surgical procedures

For implant installation, animals were anesthetized with 70 mg/kg intramuscular ketamine (Vetaset - Fort Dodge Saúde Animal Ltda) and 7 mg/kg xylazine hydrochloride (Dopaser - Laboratório Calier do Brasil Ltda) and received mepivacaine hydrochloride (0.3 ml/kg, Scandicaíne 2% with adrenaline 1:100,000; Septodont) as local anesthesia and for hemostasis of the surgical field. Antisepsis of the area to be incised was performed using polyvinyl pyrrolidone iodine degermante (PVPI 10%, Riodeine Degermante, Rioquímica). The trichotomy was performed on the medial portion of the right and left tibia. An incision approximately 1.5 cm long was made in the region of the tibial metaphysis of each tibia. Subsequently, the soft tissue was dissected to its full thickness and removed to expose the bone, which would receive the implants. Before implant installation, milling (3 mm) was performed using a spiral cutter mounted on an electric motor at a speed of 800 rpm, under irrigation with a 0.9% NaCl solution and a 20:1 reduction contra-angle. Then, the implants were installed using a digital key, and a 1 mm bone defect was created around the cervical threads of the implant [91]. Subsequently, a hydrogel was applied around the implant in the region of the bone defect corresponding to the previously in vitro contaminated cervical third. The tissues were sutured in planes with absorbable thread in the deep plane and with monofilament thread (Nylon 4.0, Ethicon, Johnson) in the outermost plane. Each animal received 2 implants, one in each tibial metaphysis. After 28 days, the animals were euthanized by anesthetic overdose, an intramuscular injection of sodium thiopental (150 mg), and the analyses were then carried out [92].

4.22. Analysis of biofilm colony-forming units and composition by 16S rRNA gene sequencing

For the biofilm analysis, the implants were removed using a digital key by counterclockwise movement. After removing the implants, they were washed in 0.9% NaCl and transferred to cryogenic tubes containing 1 mL of 0.9% NaCl. The tubes were then vortexed for 10 s and sonicated (7 W for 30 s) [93,94]. Biofilm analyses included CFU counts and sequencing; the details of these methodologies have been described above.

4.23. Analysis of the expression of inflammatory cytokines

The anti- and pro-inflammatory cytokines involved in the pathogenesis of peri-implantitis were evaluated in the in vivo model. For this analysis, after removing the implants, the tibias were reduced with a trephine, respecting at least 0.5 cm on each side of the peri-implant space to preserve the bone in contact with the implant threads. The bone fragments were collected and stored in RNAlater solution (ThermoFischer Scientific), according to the manufacturer's instructions (storage at −80 °C until use). One day before analysis, the tissues were homogenized in saline solution and centrifuged at 3000 rpm for 10 min. The supernatant was then collected and kept at 4 °C until use. Cytokine assays were performed using specific kits (Millipore Corporation, Billerica, MA, USA) with multiplex technology (Milliplex® Map Rat Cytokine). The samples were analyzed for the presence of the cytokines IL-1β, IL-4, IL-6, IL-17, TNF-α, and IFN-γ. The cytokine concentrations were determined using a standard curve of bovine serum albumin, as described by the Bradford method. The results were expressed in pg/mg by reference to the standard curve [95].

4.24. Histological analysis

Immediately after euthanasia, for histological analysis, the tibia was removed and fixed in formaldehyde solution, washed in water, and decalcified in EDTA (10%) for eight weeks. At this point, the implants were removed by counterclockwise movement with a digital key. The specimens were then dehydrated in a sequence of alcohols and diaphanized with xylene for subsequent paraffin embedding. After obtaining samples during the paraffin-embedding stage, 5-μm-thick sections were prepared and subsequently mounted on slides [92]. After microtomy, the slides were stained with hematoxylin and eosin (HE) to analyze the tissue using an optical microscope (LeicaR DMLB) associated with a camera (LeicaR DC 300F Microsystems Ltd.) [96].

4.25. Micro-CT analysis

Micro-computed tomography was carried out using a SkyScan 1176 (Bruker microCT, Aartselaar, Belgium). The region of interest (ROI) was a rectangular area, with standardized height and width, corresponding to the region of the bone defect created in the cervical portion of the implant. The analysis was carried out using the CT Analyzer (SkyScan, Leuven, Belgium). The volumetric parameters selected were bone volume (BV), bone volume percentage (BV/TV), and trabecular thickness (Tb.Th) [97].

4.26. Immunohistochemical analysis

For immunohistochemical detection of osteopontin (OPN) and osteocalcin (OCN), odd-numbered histological sections were selected. Antigen retrieval was carried out by immersing the slides in citrate buffer (Spring Bioscience, Pleasanton, CA, USA) and heating them in a pressurized decloaking chamber (Decloaking Chamber®, BioCare Medical, Concord, CA, USA) at 95 °C for 20 min. After each step of the immunohistochemical procedure, the sections were rinsed with 0.1 M phosphate-buffered saline (PBS, pH 7.4). Endogenous peroxidase activity was quenched by incubation with 3% hydrogen peroxide for 1 h, followed by blocking of nonspecific binding sites with 1% bovine serum albumin for 12 h. The sections were then incubated with goat anti-osteopontin and goat anti-osteocalcin primary antibodies (sc-21742 and sc-30044, respectively; Santa Cruz Biotechnology, Dallas, TX, USA). Subsequently, the samples were treated with a biotinylated secondary antibody for 2 h and incubated with streptavidin conjugated to horseradish peroxidase for 1 h using a universal HRP streptavidin–biotin detection system (Universal Dako Labeled HRP Streptavidin-Biotin Kit®, Dako Laboratories, CA, USA). Immunoreactivity was visualized with 3,3′-diaminobenzidine (DAB) as the chromogen (DAB Chromogen Kit®, Dako Laboratories, CA, USA). Counterstaining was performed with Harris hematoxylin, followed by dehydration in graded ethanol, clearing in xylene, and mounting with a permanent mounting medium (Permount™, Fisher Scientific, San Diego, CA, USA) and glass coverslips. Negative control sections were processed in parallel under identical conditions, except for the omission of the primary antibodies. Images were captured at 40 × magnification and subjected to semiquantitative analysis based on a scoring system. Score 0 indicated absence of immunostaining; score 1, a low immunostaining pattern (approximately 25% of positive cells); score 2, a moderate immunostaining pattern (approximately 50% of positive cells); and score 3, a high immunostaining pattern (approximately 75% of positive cells) [74].

5. Statistics

GraphPad Prism software version 9.0.0 (GraphPad, La Jolla, CA, USA) and SPSS version 26.0.0.0 (IBM Corp., Armonk, NY, USA) were used for statistical analyses. The normality of data distribution was assessed using the Shapiro-Wilk test (p < 0.05). For data with a normal distribution, a one-way ANOVA test was performed. For non-parametric data, the Kruskal-Wallis test was performed. Tukey's HSD test for multiple comparisons between groups was performed (p-value of <0.05 was used as a statistical significance threshold).

CRediT authorship contribution statement

Caroline Dini: Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Stéfany Barbosa Alves da Cruz: Formal analysis, Data curation. Rodolfo D. Piazza: Formal analysis, Data curation. Bruna E. Nagay: Methodology, Formal analysis, Data curation. Rodrigo F.C. Marques: Formal analysis, Data curation. Renato C.V. Casarin: Visualization, Formal analysis, Data curation. Fang Yang: Formal analysis, Data curation. Magda Feres: Writing – review & editing, Writing – original draft. Jet Liu: Formal analysis, Data curation. Batbileg Bor: Writing – review & editing, Writing – original draft, Visualization. Edilson Ervolino: Formal analysis, Methodology, Writing – review & editing. Liqun Xu: Formal analysis, Methodology, Writing – review & editing. Erica D. de Avila: Writing – review & editing, Writing – original draft, Visualization, Formal analysis. Leonardo P. Faverani: Writing – review & editing, Visualization, Formal analysis. João Gabriel S. Souza: Writing – review & editing, Writing – original draft, Visualization, Formal analysis. Jeroen JJP. van den Beucken: Writing – review & editing, Writing – original draft, Supervision, Methodology, Formal analysis, Conceptualization. Valentim A.R. Barão: Writing – review & editing, Writing – original draft, Supervision, Project administration, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Data availability statement

The microbiome datasets produced and analyzed in this study will be deposited in a public repository upon acceptance of the manuscript.

Ethics approval and consent to participate

This study was approved by the local Research and Ethics Committee (C.A.A.E. 49523421.0.0000.5418), the Radboud University Medical Center Ethics Committee for the use of human gingival fibroblasts (CMO Radboudumc; dossier #2017–3252), and the Animal Ethics Committee of FOA/UNESP (protocol 265-2023), following the Ethical Principles of Animal Experimentation.

Funding statement

The work was supported by The São Paulo Research Foundation (FAPESP), Grant/Award numbers: 2020/05231-4, 2020/05234-3, 2022/16267-5, 2023/02180-8; National Council for Scientific and Technological Development (CNPq) (BRICS grant number 440104/2022-0 and 307471/2021-7), Coordination of the Improvement of Higher Education Personnel (CAPES), Finance code 001. The authors thank S.I.N. Implants, São Paulo, Brazil, for their support by donating the HAnano Unitite discs and implants.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Fang Yang is an associate editor for Bioactive Materials and was not involved in the editorial review or the decision to publish this article.

Appendix A. Supplementary data

The following is the Supplementary data to this article: