Hydroxyl radicals generated by hydrogen peroxide photolysis recondition biofilm-contaminated titanium surfaces for subsequent osteoblastic cell proliferation

Titanium dental implants have been successfully used for decades; however, some implants are affected by peri-implantitis due to bacterial infection, resulting in loss of supporting bone. This study aimed to evaluate the effect of an antimicrobial chemotherapy employing H2O2 photolysis—developed to treat peri-implantitis—on biofilm-contaminated titanium surfaces in association with osteoblastic cell proliferation on the treated surface. Titanium discs were sandblasted and acid-etched, followed by contamination with a three-species biofilm composed of Porphyromonas gingivalis, Fusobacterium nucleatum, and Streptococcus mitis. This biofilm model was used as a simplified model of clinical peri-implantitis biofilm. The discs were subjected to ultrasound scaling, followed by H2O2 photolysis, wherein 365-nm LED irradiation of the disc immersed in 3% H2O2 was performed for 5 min. We analysed proliferation of mouse osteoblastic cells (MC3T3-E1) cultured on the treated discs. Compared with intact discs, biofilm contamination lowered cell proliferation on the specimen surface, whereas H2O2 photolysis recovered cell proliferation. Thus, H2O2 photolysis can recover the degraded biocompatibility of biofilm-contaminated titanium surfaces and can potentially be utilised for peri-implantitis treatment. However, to verify the findings of this study in relation to clinical settings, assessment using a more clinically relevant multi-species biofilm model is necessary.


Effect of H 2 o 2 photolysis on titanium surfaces without biofilm contamination. Microscopic
images of newly prepared titanium discs (New-Ti) treated for 5 min with LED irradiation of 3% H 2 O 2 at a wavelength of 365 nm (H 2 O 2 photolysis) denoted as H(+)L(+) and those treated with pure water without LED irradiation denoted as H(−)L(−) showed similar micro-and nano-structural attributes at the titanium surface (Fig. 1). Surface analysis using an optical interferometer and atomic force microscope (AFM) revealed that H(+)L(+) treatment did not significantly affect the micro-or nano-scale surface roughness of New-Ti (p > 0.05; Table 1).
X-ray photoelectron spectroscopy (XPS) demonstrated the presence of titanium, oxygen, and carbon on the surface of New-Ti (Fig. 2a). When titanium specimens were stored under ambient conditions for 4 weeks after specimen preparation (Aged-Ti), the percentage of carbon significantly increased from 21% (New-Ti) to 42% (Aged-Ti; p < 0.01; Fig. 2b). H(+)L(+) treatment significantly decreased the percentage of carbon on Aged-Ti to 23% (p < 0.01) but did not affect the carbon percentage of New-Ti (p > 0.05). Treatment with LED irradiation alone [H(−)L(+)] slightly decreased the percentage of carbon on Aged-Ti to 36%, but it was not significant (p = 0.08). Neither H(−)L(−) nor H(+)L(−) (immersion in 3% H 2 O 2 ; no LED irradiation) treatment significantly affected the percentage of carbon on Aged-Ti (p > 0.05).
On the surface of New-Ti, a droplet of ultrapure water (Milli-Q ® water) spread quickly indicating that the surface was hydrophilic. The contact angle of New-Ti was approximately 10° regardless of H(+)L(+) treatment (Fig. 2c). Upon aging, the surface became hydrophobic and exhibited a contact angle of >50°. H(+)L(+) treatment significantly decreased the 54° contact angle of Aged-Ti to 12° (p < 0.01), thereby recovering surface hydrophilicity.  Table S1).

Effect of ultrasound scaling followed by H 2 o 2 photolysis on
Confocal laser scanning microscopy (CLSM) of Aa biofilms with and without ultrasound scaling showed the 3D structure of each biofilm (Fig. 3a,b). Ultrasound scaling eliminated large portions of the biofilm, although some bacteria persisted on the surface as observed by CLSM (Fig. 3b). Scanning electron microscopy (SEM) revealed that the surface of Aa biofilm-Ti was entirely covered with bacterial cells and extracellular matrix (Fig. 3c,d). Furthermore, SEM showed that several bacterial cells persisted in micro-pits of the titanium rough surface after ultrasound scaling (Fig. 3e). On titanium surfaces subjected to ultrasound scaling using PEEK tips, protrusions of micro-roughened surfaces that appeared collapsed were observed in secondary electron images  Fig. 5a,b). Aging of the titanium surface significantly reduced cell proliferation (p < 0.01; Fig. 5c,d). Of the treatments tested, only H(+)L(+) significantly increased the proliferation of cells cultured on Aged-Ti as evidenced by both MTT and NR assays (p < 0.05); however, relative optical density (OD) values obtained via MTT assays (hereafter referred to as the MTT value) of H(+)L(+)-treated Aged-Ti were still significantly lower than those of cells cultured on New-Ti (p < 0.01; Fig. 5c), whereas relative OD values obtained via NR assays (hereafter referred to as the NR value) of H(+)L(+)-treated Aged-Ti were comparable to those of cells cultured on New-Ti (p > 0.05; Fig. 5d). CLSM analysis revealed that cells were densely proliferated on New-Ti after a 3-d culture irrespective of H(−)L(−) or H(+)L(+) treatment (Fig. S2) and there were no significant differences in cellular coverage and cell number between the two groups (p > 0.05). Compared with New-Ti, the cell density of H(−)L(−)-treated Aged-Ti was lower than that of New-Ti (Fig. 5e); however, cell proliferation on H(+)L(+)-treated Aged-Ti was comparable to that of New-Ti (Fig. 5e). Quantitative analysis confirmed that cellular coverage and cell number of New-Ti and H(+)L(+)-treated Aged-Ti were significantly higher than those of H(−)L(−)-treated Aged-Ti (p < 0.05), while there was no significant difference between New-Ti and H(+) L(+)-treated Aged-Ti (p > 0.05; Fig. 5f,g). Regarding cell morphometric analyses, there were no significant differences in the area (cell size), Feret's diameter, and perimeter of cells grown on New-Ti, H(−)L(−)-treated Aged-Ti, and H(+)L(+)-treated Aged-Ti (p > 0.05; Fig. S3).
Osteoblast proliferation on A. actinomycetemcomitans biofilm-contaminated titanium surfaces. Regarding CHX and PI treatment, the MTT and NR values of cells on treated Aa biofilm-Ti were significantly lower than those of cells on New-Ti (p < 0.01) and were not significantly different from those of cells on Aa biofilm-Ti treated with pure water (p > 0.05; Fig. 6c,d). In addition, treatment duration influenced the effect of H(+)L(+) treatment on the proliferation of cells on Aa biofilm-Ti (Fig. 6e,f). The longer the treatment duration, the higher the MTT and NR values.
CLSM revealed that cell density on H(−)L(−)-treated Aa biofilm-Ti after 3-d culture was sparser than that on New-Ti, whereas cell density on H(+)L(+)-treated Aa biofilm-Ti was comparable to that on New-Ti (Fig. 6g). Quantitative analysis confirmed that the number of cells on H(−)L(−)-treated Aa biofilm-Ti was significantly lower than that of New-Ti and Aa biofilm-Ti treated with H(+)L(+) (p < 0.01); however, there were no significant differences between New-Ti and Aa biofilm-Ti treated with H(+)L(+) (p > 0.05; Fig. 6h). In addition, cell morphometric analysis revealed that cells attached to H(−)L(−)-treated Aa biofilm-Ti remained small after 3-h culture, whereas those attached to New-Ti and Aa biofilm-Ti treated with H(+)L(+) became elongated (Fig. S4a). Accordingly, the area, Feret's diameter, and perimeter of cells grown on H(−)L(−)-treated Aa biofilm-Ti were significantly decreased compared with cells grown on New-Ti and Aa biofilm-Ti treated with H(+)L(+) (p < 0.05; Fig. S4b). Furthermore, the area dimensions of cells grown on H(+)L(+)-treated Aa biofilm-Ti were significantly higher than those grown on New-Ti (p < 0.01). www.nature.com/scientificreports www.nature.com/scientificreports/ Osteoblast proliferation on peri-implantitis-related three-species biofilm. To verify the effect of H 2 O 2 photolysis on cell proliferation under more clinically relevant conditions, a three-species biofilm (3S biofilm) composed of Porphyromonas gingivalis, Fusobacterium nucleatum, and Streptococcus mitis was prepared and used for cell proliferation assays. SEM images of the 3S biofilm and each bacterial species are shown in Fig. 7a-e. The entire surface of the titanium specimen was covered with many cocci followed by spindle-shaped rods and fewer short-rods, which correspond to S. mitis, F. nucleatum, and P. gingivalis, respectively. Quantitative real-time polymerase chain reaction (qPCR) analysis demonstrated that the colony forming equivalents (CFE) of P. gingivalis, F. nucleatum, and S. mitis in the 3S biofilm were 7.1 × 10 2 , 5.0 × 10 6 , and 2.2 × 10 7 CFE/specimen, www.nature.com/scientificreports www.nature.com/scientificreports/ respectively (Fig. 7f). Total viable counts evaluated by the culture method were 4.4 × 10 6 CFU/specimen (Fig. 7g), which was lower than the total of each bacterial CFE/specimen. Ultrasound scaling resulted in significant reduction of CFE/specimen for all three bacterial species (P. gingivalis, not detected; F. nucleatum, 6.0 × 10 2 CFE/ specimen; and S. mitis, 1.4 × 10 4 CFE/specimen; Fig. 7f). Although bacterial DNA was detected by qPCR after ultrasound scaling, viable bacteria were not detected through culturing, regardless of bacterial species (Fig. 7g).

Discussion
The present study showed that H 2 O 2 photolysis can recover not only aged titanium surfaces but also biofilm-contaminated ones that cannot be completely cleaned by ultrasound scaling, thereby obtaining like-new conditions suitable for osteoblastic cell proliferation. Consequently, H 2 O 2 photolysis enabled the MC3T3-E1 osteoblastic cells to proliferate on Aged-Ti, Aa biofilm-Ti, and 3S biofilm-Ti to the same extent as proliferation on New-Ti. Furthermore, this treatment did not affect the physical and chemical properties of intact titanium surfaces nor the associated cell response. Therefore, our results suggest that photolysis of 3% H 2 O 2 with 365-nm LED irradiation can be utilised for chemical decontamination of aged and/or biofilm-contaminated titanium surfaces without material damage.
For clinically relevant findings, the effect of antimicrobial agents used for peri-implantitis treatment should be studied using titanium specimens with rough surfaces contaminated by microbial biofilms and the assessment of surface decontamination should involve quantification of residual biofilms 20 . The present study was performed in accordance with these recommendations. Titanium specimen surfaces were prepared to replicate the rough surfaces of dental implants, namely the sandblasted, large grit, and acid-etched (SLA) surfaces 27 . Analysis of  www.nature.com/scientificreports www.nature.com/scientificreports/ surface roughness at the micro-and nano-scales confirmed that the specimens used herein had similar surface roughness to SLA surfaces 28 . Regarding surface contamination, Aa biofilms were used because clinical studies indicated the possible association of this bacterial species with peri-implantitis [29][30][31][32] . To mimic clinical procedures for implant surface debridement, ultrasound scaling using a PEEK tip was performed so as to not damage the rough surface of titanium in accordance with previous studies 33,34 , followed by treatment with each antimicrobial agent. Subsequently, biofilm-forming bacteria were quantified via colony enumeration assays after each treatment and the proliferation of osteoblastic cells on the treated titanium surface was evaluated. Cell proliferation was analysed via MTT and NR assays in the present study. MTT assays assess mitochondrial succinate dehydrogenase activity in viable cells 35 , whereas NR assays assess lysosomal incorporation of the dye in viable cells 36 . Although both measures are essentially correlated with the number of viable cells, they do not exactly reflect cell viability, especially when the function and/or number of mitochondria and lysosomes are affected by drug treatments 37 . Thus, both assays were used and the results were further confirmed via CLSM analysis.
The mechanism underlying 'biological aging of titanium' caused by storage under ambient conditions has been well documented by Att et al. 38 . They reported that aging is characterised by time-dependent accumulation of hydrocarbons on the titanium surface. Although newly prepared titanium surfaces are hydrophilic due to the affinity of the positively charged TiO 2 layer at the outermost titanium surface for polar water molecules, accumulation of non-polar hydrocarbons makes the surface hydrophobic 39 . In addition, loss of the electropositive surface is not favourable for cellular attachment because cells and proteins associated with cell adhesion are negatively charged 39,40 . Att et al. [38][39][40][41][42] , who reported the aging of titanium, conducted extensive research on this issue, and devised a technique to recover the lost bioactivity of aged titanium surfaces, which was termed 'photo-functionalization' . Photo-functionalization is performed using ultraviolet (UV) irradiation, which decreases the amount of hydrocarbons on the surface of aged titanium, converting the hydrophobic surface to a hydrophilic one. On the UV-treated surfaces of aged titanium, osteoblasts showed enhanced proliferation when compared with those on untreated surfaces of aged titanium. In the present study, H 2 O 2 photolysis [H(+)L(+)] induced analogous chemical and biological changes. By contrast, LED irradiation at 365 nm without H 2 O 2 [H(−) L(+)] did not significantly influence carbon reduction and cell proliferation on aged titanium surfaces, suggesting that H 2 O 2 photolysis exerts its effect through a different mechanism than that of photo-functionalization. For photo-functionalization, aged titanium is generally irradiated with UVC (250 nm) at 2 mW/cm 2 for 48 h [40][41][42] . Thus, the radiant exposure is calculated to be 345.6 J/cm 2 (0.002 W/cm 2 × 172,800 s). Herein, the specimens were irradiated with UVA (365 nm) at 1000 mW/cm 2 for 5 min, corresponding to a radiant exposure of 300 J/cm 2 (1 W/cm 2 × 300 s). Although the radiant exposure provided was comparable, UVA irradiation did not result in photo-functionalization, suggesting that this effect depends on the wavelength of light, concurrent with previous reports wherein UVA treatment of aged titanium displayed no effects on osteoblast attachment and proliferation 41,42 . UVC irradiation has been suggested to eliminate carbon on titanium surfaces through two mechanisms, photocatalytic activity of the TiO 2 layer and direct decomposition of carbon compounds by UVC 42 . UVA can also induce photocatalytic reactions with TiO 2 . The photon energy of UVA at 365 nm-calculated by the equation of E = hc/λ, where E is energy, h is Planck's constant, c is the speed of light, and λ is the wavelength-is 3.4 eV, which is larger than the band gap of TiO 2 (3.2 eV for anatase) 43 . Moreover, the 365-nm LED irradiation of titanium specimens immersed in pure water resulted in more than double the yield of hydroxyl radicals compared with irradiation of pure water without titanium specimens, as evidenced by ESR analysis and probably owing to the photocatalytic activity of TiO 2 . This finding suggests that hydroxyl radicals generated by UVA-induced photocatalysis are not sufficient for reducing carbon levels on aged titanium surfaces as shown in Fig. 2b. Therefore, H 2 O 2 photolysis, which generates more hydroxyl radicals than photocatalysis, possibly results in the decomposition of accumulated carbon compounds on aged titanium surfaces.
When assessing cell proliferation on Aged-Ti, MTT and NR assays showed similar results regarding H(+) L(+) treatment in recovering the aging-induced reduction in cell proliferation; however, there were some discrepancies. For instance, although MTT assay results suggested that H(+)L(+) treatment accelerated cell proliferation on Aged-Ti, this proliferation was still much lower than cell proliferation on New-Ti. On the other hand, NR assays showed comparable cell viability between H(+)L(+)-treated Aged-Ti and New-Ti, whereas the other treatments resulted in significantly decreased cell viability compared with that of New-Ti. The discrepancy between MTT and NR assays probably results from differences in assay conditions, as described above. CLSM analysis demonstrated that the number of cells on H(+)L(+)-treated Aged-Ti was equivalent to that on New-Ti, whereas the number of cells on H(−)L(−)-treated Aged-Ti was approximately half in comparison. Considering this, it was determined that H 2 O 2 photolysis can recover the surface condition of aged titanium for subsequent osteoblast proliferation to achieve the same conditions as newly prepared materials. However, the activity and number of mitochondria and/or lysosomes in osteoblastic cells growing on aged titanium surfaces treated with H 2 O 2 photolysis may be affected via modified surface properties, resulting in the observed discrepancy between MTT and NR assays. We intend to further investigate whether this discrepancy affects the establishment of osseointegration in future studies.
Parameter values for cell proliferation on Aa biofilm-Ti were significantly lower than those for cells grown on New-Ti. However, H(+)L(+) treatment recovered the reduction in cell proliferation induced by Aa biofilm contamination. Values and error bars in the graphs indicate the mean and standard deviation, respectively [n = 6 for (a-f) and n = 9 for (h)]. Different letters above the columns in the graphs refer to significant differences (p < 0 www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ Biofilm contamination assays confirmed that Aa biofilms on titanium surfaces cannot be detached from the surface by simply washing with saline. CLSM and SEM confirmed that the biofilm was comprised of bacterial cells and extracellular matrix components. The initial viable count of 1.3 × 10 5 CFU/specimen in the biofilm was reduced to 1.5 × 10 2 CFU/specimen when subjected to ultrasound scaling. Although ultrasound scaling effectively eliminated microbial biofilms, it could not completely eliminate bacteria, as evidenced by colony counting assays, CLSM, and SEM, which is concurrent with a previous report 33 . Moreover, SEM revealed that protrusions of micro-roughened titanium surfaces appeared collapsed in some parts after ultrasound scaling. In the backscattered electron image, a clear contrast was observed between the intact titanium surface and protrusions. Since the yield of backscattered electrons depends on the constituent element, the dark image at the protrusions indicates the presence of an element other than titanium with a lower atomic number. Thus, protrusions of micro-roughened titanium surfaces were plugged by remnants of the PEEK tip rather than being mechanically damaged. This was also confirmed via XPS wherein increased carbon compounds on New-Ti were observed after ultrasound scaling. Bacteria left on the titanium surface after ultrasound scaling were subjected to additional antimicrobial treatments. All antimicrobial treatments tested herein reduced the viable counts below the detection limit (2 CFU) after 5 min of treatment. Because viable bacterial cell counts were reduced via ultrasound scaling prior to antimicrobial treatment, differences in the bactericidal effect between treatments could not be detected. Therefore, cell proliferation assays following antimicrobial treatments may be barely affected by viable bacteria and solely influenced by alterations in surface conditions caused by biofilm contamination and treatments.
XPS demonstrated that ultrasound scaling of Aa biofilm-Ti yielded a higher atomic percentage of carbon than did New-Ti. Furthermore, Aa biofilm-Ti exhibited a nitrogen peak that was not present for New-Ti. These findings confirmed that biofilm contamination increased the percentage of both carbon and nitrogen on the titanium surface as reported previously 44 . Although H(+)L(−) treatment did not affect the percentage of carbon, H(−) L(+) and H(+)L(+) significantly decreased the percentage of carbon on Aa biofilm-Ti. The effect of H(−)L(+) is probably attributable to photocatalytic action of the TiO 2 layer formed on specimen surfaces, although the same treatment did not significantly decrease the atomic percentage of carbon accumulated on Aged-Ti. This may be due to the differing affinities of different types of carbon compounds derived from biofilm contamination and aging. As H 2 O 2 photolysis can more effectively reduce the carbon on Aa biofilm-Ti than can H(−)L(+), H 2 O 2 photolysis can thus effectively reduce carbon compounds regardless of origin.
Furthermore, H 2 O 2 photolysis was able to recondition Aa biofilm-Ti surfaces for subsequent osteoblastic cell proliferation. MTT assays revealed that cell proliferation on H(−)L(−)-treated Aa biofilm-Ti was decreased compared with that on New-Ti. Conversely, cell proliferation on H(+)L(+)-treated Aa biofilm-Ti increased compared with that on New-Ti. Furthermore, H(−)L(+)-treated Aa biofilm-Ti yielded comparable MTT assay results to that of New-Ti. Similar tendencies were observed for NR assays. Comparison of H(+)L(+) and H(−)L(+) suggests that the former reconditions titanium surfaces contaminated with microbial biofilms more effectively for subsequent osteoblastic cell proliferation than can the latter. Considering XPS and cell proliferation assay results, this accelerated cell proliferation is possibly associated with elimination of carbon contamination by each treatment. Additionally, the effect of H(+)L(+) was enhanced in a time-dependent manner. Of the conditions tested, 5-min H 2 O 2 photolysis exerted the highest effect on cell proliferation. In contrast, treatment with antiseptics, such as 3% H 2 O 2 alone, 0.5% PI, and 0.2% CHX, exerted minor effects on subsequent cell proliferation on Aa biofilm-Ti, although all treatments tested reduced viable bacterial counts to below the detection limit. CLSM also confirmed the H 2 O 2 photolysis treatment effects observed in the MTT and NR assays. Osteoblastic cells attached to Aa biofilm-Ti and cultured for 3 h were less elongated compared with those on New-Ti. However, elongation of cells attached to H 2 O 2 photolysis-treated Aa biofilm-Ti were comparable to or more than those on New-Ti after 3-h culture. Initial cell-biomaterial interactions, such as cell attachment and spreading, generally regulate subsequent function 45,46 . For instance, cell spreading enhances osteoblast differentiation and osteoblast matrix deposition 47,48 . Unsuccessful elongation of osteogenic cells may result in little induction of osteogenic phenotypes and may lead to their de-differentiation 40,49 . Thus, the cell spreading on H 2 O 2 photolysis-treated Aa biofilm-Ti observed in the present study indicates that this treatment can re-establish the surface for favourable cell-biomaterial interactions. Additionally, after 3-d culture, the cell count on H 2 O 2 photolysis-treated Aa biofilm-Ti was equivalent to that on New-Ti, whereas the number of cells on H(−)L(−)-treated Aa biofilm-Ti was significantly decreased compared with those on New-Ti and H 2 O 2 photolysis-treated Aa biofilm-Ti.
To verify the findings obtained using the Aa biofilm model, 3S biofilm-Ti-a more clinically relevant modelwas prepared using the peri-implantitis-related bacteria P. gingivalis, F. nucleatum, and S. mitis 50 in basal mucin medium, which is frequently used as artificial saliva [51][52][53] , and subjected to MTT assays. Composition of the 3S biofilm was similar to a previous report 53 , whereby S. mitis was the most dominant species followed by F. nucleatum and then P. gingivalis in small amounts. The finding that total CFE/specimen was higher than total CFU/ specimen indicates that the biofilm contained a substantial number of dead bacterial cells. Although ultrasound scaling resulted in no detection of viable bacteria, the titanium surface still contained dead bacterial cells and/ or extracellular bacterial DNA as demonstrated by qPCR. Thus, as in the case of the Aa biofilm, the reduced cell proliferation on H(−)L(−)-treated 3S biofilm-Ti compared with New-Ti is possibly due to contaminants such H(−)L(+). Values and error bars in the graphs indicate the mean and standard deviation, respectively [n = 4 for (f,g) and n = 6 for (h,i)]. **Significant differences (p < 0.01) in CFE/specimen of each bacterial species or total viable counts between US(−) and US(+) groups in (f,g). Different letters above the columns in (h,i) refer to significant differences (p < 0.05) between different groups. ND In addition, it was demonstrated that even 1-min H(+)L(+) treatments recovered cell proliferation to comparable levels as on New-Ti, indicating that treatment time may be shortened to less than 5 min. Considering the cell proliferation assay results of both Aa and 3S biofilm models, osteoblastic cells are able to proliferate on biofilm-contaminated titanium surfaces after H 2 O 2 photolysis treatment and cell proliferation can be recovered to the same levels as on New-Ti.
Nevertheless, the present study has some limitations. For instance, only biofilm models composed of limited bacterial species were used to simplify the study for evaluation of various testing conditions, even though clinical biofilms at peri-implantitis sites have a much larger diversity composed of >150 microbial species, as demonstrated by recent metagenomics studies 54,55 . In addition, it should be noted that although they are associated with periodontitis, the bacterial species used in the present study, especially A. actinomycetemcomitans, may not represent peri-implantitis pathogens as the microbiomes of periodontal and peri-implant biofilm microbiomes may be distinct [56][57][58] . Accordingly, the findings of the present study should be validated using more clinically relevant models, such as an ex vivo biofilm model (using microorganisms collected from peri-implantitis sites) or an in vivo peri-implantitis model. Another limitation is that the mouse osteoblastic cell line was used instead of cells originating from humans, and bone formation on biofilm-contaminated titanium surfaces was not evaluated. Therefore, more extensive studies are required to verify efficacy of the H 2 O 2 photolysis technique in peri-implantitis treatment.
Within the limitations of the present study, we concluded that H 2 O 2 photolysis can eliminate accumulated carbon compounds on titanium surfaces due to both aging and microbial biofilm contamination as well as enhance subsequent osteoblastic cell proliferation. In contrast, conventional antiseptics, such as PI, CHX, and H 2 O 2 (without LED irradiation), did not recover cell proliferation on biofilm-contaminated titanium surfaces. Although the effects of H 2 O 2 photolysis and photo-functionalization are similar, it may be difficult to apply the latter treatment to peri-implantitis because UVC irradiation can damage the DNA of host tissue cells 59 . As for the safety of H 2 O 2 photolysis, 3% H 2 O 2 is regarded as acceptable 60 and UVA has also been employed in photo-therapy for the treatment of skin conditions 61,62 . In addition, exposure of the oral mucosa to hydroxyl radicals for a short period of time does not cause abnormal changes, as reported in previous in vivo studies 63,64 . Although safety assessment of this treatment-5-min UVA irradiation of 3% H 2 O 2 on the lesion site-should be conducted before clinical application, H 2 O 2 photolysis has potential for use as a peri-implantitis treatment by inactivating microbial biofilms and reconditioning titanium surfaces for subsequent osteoblast proliferation with regard to efficacy. www.nature.com/scientificreports www.nature.com/scientificreports/

Materials and Methods
Preparation of titanium specimens. Commercially pure titanium discs (Grade 4; TB550; Nishimura, Fukui, Japan) with 5-mm diameters and 2-mm thick were sandblasted and acid-etched to obtain a rough surface similar to that of commercial dental implants, namely SLA surfaces. Sandblasting was performed using 250-µm alumina particles at a blasting pressure of 0.4 MPa. The specimens were then acid etched in 49% (v/v) sulfuric acid for 1 h at 60 °C and cleaned via ultrasonic cleaning in ultrapure water (10 min), acetone (10 min), and again in ultrapure water (10 min). After cleaning, the specimens were autoclaved for 15 min at 121 °C. Acid etching, cleaning, and autoclaving were performed the day before each assay, unless otherwise specified. Analogously, L(+) specimens were irradiated with 365-nm LED light at 1000 mW/cm 2 , whereas L(−) specimens were maintained in a light-shielding box. We used an LED spot-curing device (OmniCure LX400; Lumen Dynamics Group, Mississauga, Canada) as the light source. The specimens were washed twice with ultrapure water and assayed.

Assessment of H
Surface topography of newly prepared titanium discs (New-Ti) treated with H(−)L(−) and H(+)L(+) was examined using an SEM (JSM-7100F; JEOL, Tokyo, Japan), an optical interferometer (Talysurf CCI HD; Taylor Hobson, Leicester, UK), and an AFM (SFT-3500; Shimadzu, Kyoto, Japan). The SEM was operated at 5 kV, the optical interferometer under a 50 × objective lens to analyse surface roughness at the micro-scale, and the AFM in tapping mode using a silicon cantilever (spring constant, 2 N/m; resonant frequency, 70 kHz) and with a scan range of 1 × 1 µm for analysing surface roughness at the nano-scale. Six independent assays were performed for each test.
The crystalline structures of New-Ti treated with H(−)L(−) or H(+)L(+) were analysed using a θ-2θ XRD (SmartLab; Rigaku, Tokyo, Japan). Diffractograms were obtained from 20° to 65° at a scan speed of 10°/min and a step size of 0.01° using Cu-Kα radiation. A high-sensitive 2D detector (HyPix-400; Rigaku) was used to detect a thin layer of TiO 2 (i.e. passive film). Diffractogram peaks were qualified using standard models recorded in the International Centre for Diffraction Data and TiO 2 peak intensities were computed using the software package PDXL (Rigaku). Four independent assays were performed for each test.
Surface chemical composition was examined using an XPS (JPS-9010MC; JEOL). New-Ti as well as titanium discs aseptically stored in an incubator at 37 °C for 4 weeks (Aged-Ti) were included in this analysis. New were subjected to XPS. The XPS device was operated at 10 kV and 10 mA as well as monochromatic Mg-Kα X-ray radiation with a spot size of 1 mm. The results were evaluated using the software package SpecXPS (JEOL) and four independent assays were performed for each test.
The wettability of New-Ti and Aged-Ti treated with H(−)L(−) and H(+)L(+) was evaluated using a contact angle meter (CA-X, Kyowa Interface Science, Saitama, Japan). The contact angle of 0.4 µL ultrapure water was measured and six independent assays were performed for each test.
A.actinomycetemcomitans biofilm contamination. The test strain A. actinomycetemcomitans JCM 2434 was obtained from the Japan Collection of Microorganisms (RIKEN BioResource Center, Wako, Japan). The study design is illustrated in Fig. S5 and biofilm formation was induced using a previously reported method 65 . The single-species biofilm model, instead of a multi-species biofilm, was used as a simplified model to evaluate various testing conditions as described below. Bacteria were grown anaerobically using AneroPack (Mitsubishi Gas Chemical Company, Tokyo, Japan) in brain heart infusion (BHI) broth (Becton Dickinson, Franklin Lakes, NJ) supplemented with 10 g/L yeast extract (Oxoid, Hampshire, UK) at 37 °C for 24 h. From this culture, a bacterial suspension was prepared in sterile saline and the concentration was adjusted to approximately 5 × 10 8 CFU/mL. Autoclaved titanium discs were placed in 48-well plates, and then 1 mL BHI broth supplemented with 10 g/L yeast extract and 100 µL bacterial suspension (inoculum = 5 × 10 7 CFU) were added to the wells, after which the plates were incubated under anaerobic conditions at 37 °C for 48 h. Thereafter, the specimens were gently washed twice with saline to eliminate non-attached bacteria and used in subsequent assays as Aa biofilm-Ti. Assessment of ultrasound scaling followed by H 2 o 2 photolysis on A. actinomycetemcomitans biofilm-contaminated titanium. Aa biofilm-Ti was subjected to ultrasound scaling (mechanical removal of biofilm), followed by each antimicrobial treatment to mimic the clinical procedure of peri-implantitis treatment. Ultrasound scaling was performed using a power-driven scaler (miniMaster Piezon LED, EMS, Nyon, Switzerland) with a plastic scaler tip made of PEEK (ITMS Peek tip straight, Star Chip, Himeji, Japan). Instrument power was set to 70% and pure water at a flow rate of 50 mL/min was used as water coolant. The titanium specimen was held using forceps and the entire biofilm-containing surface was subjected to ultrasound scaling for 1 min. After ultrasound scaling, the specimens were treated with either H ( Aa biofilm-Ti specimens subjected to each treatment were washed twice with saline to stop the reactions. Thereafter, the remaining bacteria in the biofilm were harvested via the enzymatic detachment method as previously described but with some modifications 66 . Each treated Aa biofilm-Ti was immersed in 300 µL enzyme (2019) 9:4688 | https://doi.org/10.1038/s41598-019-41126-z www.nature.com/scientificreports www.nature.com/scientificreports/ suspension composed of 4 mg/mL type I collagenase (Thermo Fisher Scientific, Waltham, MA) and 2 mg/mL dispase (Thermo Fisher Scientific) in phosphate-buffered saline (PBS). The 48-well plate containing samples was incubated for 2 h under rotation at 200 rpm and 37 °C. Subsequently, the sample and enzyme solution were transferred to 1.5-mL micro-tubes and vortexed for 10 s. The suspension was serially diluted 10-fold in saline and then 150 µL of the dilution was plated onto BHI agar (Becton Dickinson). Agar plates were cultured anaerobically at 37 °C for 48 h, followed by colony counting to determine CFU/specimen. Viable bacterial counts for Aa biofilm-Ti without any treatments were also evaluated. Three independent assays were performed for each test.
Aa biofilm-Ti with or without ultrasound scaling was analysed by CLSM and SEM. For CLSM, the biofilm samples were fluorescently stained with 10 µM SYTO9 (Thermo Fisher Scientific) for 20 min at room temperature (20-25 °C). Thereafter, samples immersed in pure water were imaged using a laser at an excitation wavelength of 488 nm with a CLSM (TCS-SPE; Leica, Wetzlar, Germany). For SEM, Aa biofilm-Ti was fixed with 2.5% glutaraldehyde, post-fixed with 2% osmium tetroxide, dehydrated through a graded ethanol series, and freeze-dried in t-butyl alcohol. Each freeze-dried sample was coated with platinum and then secondary electron images of the sample were obtained using an SEM (JSM-7100F) operated at 5 kV. In addition, backscattered electron images were also obtained using the SEM operated at 15 kV to examine the presence of PEEK tip remnants on the titanium surface after ultrasound scaling.
The surface chemical composition of New-Ti treated with or without ultrasound scaling and Aa biofilm-Ti treated with ultrasound scaling followed by H(−)L(−), H(+)L(−), H(−)L(+), or H(+)L(+) was examined using an XPS under the same aforementioned conditions. Three independent assays were performed for each test.
Cell L(+), CHX, or PI treatment were also examined. Furthermore, to evaluate the effect of treatment duration, Aa biofilm-Ti subjected to ultrasound scaling followed by H(+)L(+) treatment for 0, 1, 3, and 5 min was also subjected to the assays. For the MTT assay, MTT (Tokyo Chemical Industry, Tokyo, Japan) was dissolved in PBS and diluted in the medium for a final 0.1% (w/v) solution. The titanium specimens on which cells proliferated for 3 d (as described above) were immersed in 300 µL MTT solution and incubated for 2 h. Thereafter, MTT that was converted into insoluble formazan was solubilised using dimethyl sulfoxide and evaluated through colourimetric determination at 595 nm with a microplate reader (FilterMax F5; Molecular Devices, Sunnyvale, CA). For the NR assay, the titanium specimens were immersed in 300 µL of medium containing 150 µg/mL NR (Wako Pure Chemicals) and incubated for 3 h. Thereafter, NR incorporated into the cells was extracted with 50% (v/v) ethanol containing 1% (v/v) acetic acid for colourimetric determination at 540 nm using the microplate reader. Six independent assays were performed for each test.
For CLSM, cells grown on New-Ti, Aged-Ti treated with H(−)L(−) or H(+)L(+), and Aa biofilm-Ti treated with H(−)L(−) or H(+)L(+) were fixed in 4% paraformaldehyde and fluorescently stained. Nuclei and actin filaments were stained with 300 nM 4' ,6-diamidino-2-phenylindole (DAPI; Thermo Fisher Scientific) and 165 nM rhodamine phalloidin (Thermo Fisher Scientific), respectively. To quantify the cells on Aa biofilm-Ti, nuclei were stained with 10 µM SYTO9 instead of DAPI as the latter increases background noise. The cells were imaged using the CLSM (TCS-SPE) at an excitation wavelength of 405 nm for DAPI, 488 nm for SYTO9, and 532 nm for rhodamine phalloidin. The obtained CLSM images were analysed using ImageJ (National Institute of Health, Bethesda, MD). Cells were quantified over nine CLSM images while cell morphometry was performed across 15 images.
Preparation and characterisation of the three-species biofilm. The test strains P. gingivalis JCM 12257, F. nucleatum JCM 12990, and S. mitis JCM12971 were obtained from the Japan Collection of Microorganisms (RIKEN BioResource Center). P. gingivalis and F. nucleatum were grown anaerobically in BHI broth supplemented with 5 mg/L hemin (Tokyo Chemical Industry) and 1 mg/L menadione (Tokyo Chemical Industry) at 37 °C for 2 d. S. mitis was anaerobically grown in BHI broth for 20 h. From these cultures, bacterial suspensions of P. gingivalis and S. mitis were prepared in sterile saline while F. nucleatum was prepared in liquid Amies transport medium [3 g/L sodium chloride (Wako Pure Chemicals Industries), 1.15 g/L sodium hydrogen phosphate (Nacalai Tesque), 1 g/L sodium thioglycollate (Nacalai Tesque), 0.2 g/L potassium dihydrogen phosphate (Wako Pure Chemicals Industries), and 0.2 g/L potassium chloride (Wako Pure Chemicals Industries) in pure water]. The concentration of each bacterial suspension was adjusted to approximately 5 × 10 8 CFU/mL. For biofilm formation, basal mucin medium 51-53 was used which contains 2.5 g/L mucin from porcine stomach (Wako Pure Chemicals Industries), 3.5 g/L sodium chloride, 0.2 g/L potassium chloride, 0.2 g/L calcium chloride dihydrate (Wako Pure Chemicals Industries), 2 g/L yeast extract, 1 g/L lab-lemco powder (Oxoid), 5 g/L proteose peptone (Oxoid), 5 mg/L hemin, and 1 mg/L menadione in pure water. After autoclaving, urea (Wako Pure www.nature.com/scientificreports www.nature.com/scientificreports/ Chemicals Industries) was added for a final 0.5 g/L concentration. The autoclaved titanium discs were placed in 48-well plates, and then 820 µL basal mucin medium and 60 µL of each bacterial suspension (inoculum = 3 × 10 7 CFU) were added to each well. Subsequently, the plates were incubated under anaerobic conditions at 37 °C for 48 h. Thereafter, the specimens were gently washed twice with saline to eliminate non-attached bacteria and used in the following assays as 3S biofilm-Ti.
3S biofilm-Ti as well as single species biofilms composed of each bacterial species were subjected to SEM analysis as described above. Furthermore, 3S biofilm composition with and without ultrasound scaling was analysed using qPCR. The biofilm-forming bacteria were detached by the aforementioned enzymatic method. Subsequently, bacterial DNA was extracted using PureLink Microbiome DNA Purification Kit (Thermo Fisher Scientific). Extracted DNA (2 µL) was added to a master mix of 10 µL SYBR Green (SSoAdvanced Universal SYBR Green Supermix, Bio-Rad Laboratories, Hercules, CA), 2 µL each of 5 µM forward and reverse primers specific for each bacterial species [68][69][70] (Table S2), and 4 µL nuclease-free water. The reaction mixture was then subjected to qPCR using the MyiQ Single-Color Real-Time PCR Detection System (Bio-Rad Laboratories). To construct standard curves for each bacterial species, suspensions prepared in a 10 3 -10 8 CFU/mL range were also subjected to qPCR. Based on the standard curves, CFE/specimen was calculated. In addition, total viable counts for 3S biofilms were evaluated by the culture method. Suspensions of detached biofilm-forming bacteria were serially diluted 10-fold in saline and then 50 µL of the dilution was plated onto Brucella HK agar (Kyokuto Pharmaceutical Industry, Tokyo, Japan) supplemented with 5% (v/v) defibrinated horse blood (Nippon Bio-Supp. Center, Tokyo, Japan). Agar plates were cultured anaerobically at 37 °C for 7 d, followed by colony counting for determination of CFU/specimen. Viable bacterial counts for 3S biofilm-Ti subjected to ultrasound scaling were also evaluated. Four independent assays were performed for each test. treatments generated only trace levels of hydroxyl radicals, and thus the treatment duration was extended to 5 min. Furthermore, the yield of hydroxyl radicals generated by H(−)L(+) treatment with or without titanium specimens was compared to estimate the photocatalytic activity of TiO 2 formed on the titanium surface. ESR spectra of the samples were recorded using an X-band ESR spectrometer (JES-FA-100; JEOL). The concentration of hydroxyl radicals trapped by DMPO (DMPO-OH) was determined using the software package Digital Data Processing (JEOL). Three independent assays were performed for each test.
Statistical analysis. Differences in surface roughness, atomic percentages of carbon, contact angles, bacterial cell count, cell proliferation parameters, and hydroxyl radical yield were assessed using the Student t-test for pairwise comparisons or analysis of variance (ANOVA) followed by the post-hoc Tukey-Kramer Honestly Significant Difference test for multiple comparisons. Statistical analyses were performed using JMP Pro 13.0 software (SAS Institute, Cary, NC). p values < 0.05 were considered statistically significant.

Data Availability
The datasets generated and/or analysed during the current study are available from the corresponding author upon reasonable request.