Antimicrobial Photoinactivation Using Visible Light Plus Water-Filtered Infrared-A (VIS + wIRA) and Hypericum Perforatum Modifies In Situ Oral Biofilms

Due to increasing antibiotic resistance, the application of antimicrobial photodynamic therapy (aPDT) is gaining increasing popularity in dentistry. The aim of this study was to investigate the antimicrobial effects of aPDT using visible light (VIS) and water-filtered infrared-A (wIRA) in combination with a Hypericum perforatum extract on in situ oral biofilms. The chemical composition of H. perforatum extract was analyzed using ultra-high-performance liquid chromatography coupled with high resolution mass spectrometry (UPLC-HRMS). To obtain initial and mature oral biofilms in situ, intraoral devices with fixed bovine enamel slabs (BES) were carried by six healthy volunteers for two hours and three days, respectively. The ex situ exposure of biofilms to VIS + wIRA (200 mWcm−2) and H. perforatum (32 mg ml−1, non-rinsed or rinsed prior to aPDT after 2-min preincubation) lasted for five minutes. Biofilm treatment with 0.2% chlorhexidine gluconate solution (CHX) served as a positive control, while untreated biofilms served as a negative control. The colony-forming units (CFU) of the aPDT-treated biofilms were quantified, and the surviving microorganisms were identified using MALDI-TOF biochemical tests as well as 16 S rDNA-sequencing. We could show that the H. perforatum extract had significant photoactivation potential at a concentration of 32 mg ml−1. When aPDT was carried out in the presence of H. perforatum, all biofilms (100%) were completely eradicated (p = 0.0001). When H. perforatum was rinsed off prior to aPDT, more than 92% of the initial viable bacterial count and 13% of the mature oral biofilm were killed. Overall, the microbial composition in initial and mature biofilms was substantially altered after aPDT, inducing a shift in the synthesis of the microbial community. In conclusion, H. perforatum-mediated aPDT using VIS + wIRA interferes with oral biofilms, resulting in their elimination or the substantial alteration of microbial diversity and richness. The present results support the evaluation of H. perforatum-mediated aPDT for the adjunctive treatment of biofilm-associated oral diseases.


Results
chemical composition of the H. perforatum extract. A rapid and accurate method was developed, aiming to detect the major active compounds in the H. perforatum extract, as described elsewhere 30 . For the LC-HRMS, an Orbitrap high resolution mass analyzer was used in negative ionization mode. Peaks were identified by comparing the retention times and HRMS data of the compounds with already-published data 31,32 and natural product databases (PubChem, ChemSpider). The extract of H. perforatum contains a wide range of constituents (Table 1, Fig. 3). All of the detected compounds along with their names, Rt (min), molecular formula, theoretical and experimental m/z, mass difference (ppm) and RDBeq (Relative Double Bond equivalent) value are summarized in Table 1, whereby a total of sixteen secondary compounds were detected. Phenolic aPDT effects on the viable counts of oral microorganisms at initial adhesion and in mature oral biofilms. Figure 4a shows the effects of aPDT using VIS + wIRA and H. perforatum on the log counts of the initial and mature oral biofilms after two hours and three days, respectively. The aPDT induced a complete eradication of the bacteria (p = 0.0001, Fig. 4a) within the initial biofilms treated with non-rinsed H. perforatum, whether followed or not by aPDT, as well as within the CHX-treated initial biofilms. Interestingly, the initial biofilms in which H. perforatum was rinsed prior to aPDT exhibited a mean of 0.69 ± 0.95 CFU in the log 10 scale Figure 2. Graphical scheme depicting aPDT using VIS + wIRA as a light source and H. perforatum as a photosensitizer against oral biofilms. The VIS + wIRA with a broadband water-filtered spectrum (570-1400 nm) facilitated the excitation of the photosensitizing agent (H. perforatum), triggering its interaction with oxygen (O 2 ). ROS were then released, eradicating both planktonic and adherent oral microorganisms.

Experimental Theoretical
Δm (median, 0.25), which is a significantly reduced (p = 0.002) microbial growth compared with the untreated negative control (mean, 3.81 ± 0.66; median, 3.75). Figure 4(b,c) displays the microbial growth rates of the three-day-old oral biofilm after the application of aPDT. In the presence of H. perforatum, the aerobic and anaerobic microorganisms within the aPDT-treated biofilms showed no microbial growth and thus a CFU reduction of 100% (p = 0.0001). With regard to the aerobic biofilm bacteria, aPDT following rinsing of H. perforatum (mean, 5.49 ± 0.55; median, 5.52) yielded no significant CFU decline compared with the untreated negative control (mean, 6.23 ± 0.38; median, 6.17) and the CHX-treated positive control (mean, 5.44 ± 0.71; median, 5.45). Regarding the anaerobic biofilm bacteria, the application of aPDT after rinsing H. perforatum (mean, 4.49 ± 1.00; median, 4.64) failed to yield a decrease in the log 10  Shift of bacterial spectrum after the use of apDt against initial microbial adhesion. The composition of cultivable bacteria in initial adhesion samples is analyzed in Figs. 5A,B, 6A, and 7A,B. After treatment with CHX (positive control) and H. perforatum, whether followed or not by aPDT, all bacteria were eliminated. From the original bacterial community in the untreated negative control, a total of 31 different cultivable species were isolated and identified, as shown in Fig. 5A. The microbial community was predominated (29%) by eight different streptococcal species (Streptococcus mitis/oralis 11%; Streptococcus sanguinis/parasanguinis, 7%; Streptococcus salivarius, 4%; Streptococcus australis, 3%; Streptococcus infantis, 3%; Streptococcus gordonii, 1%). Different species of the genera Rothia (Rothia dentocariosa, Rothia aeria, Rothia mucilaginosa), Actinomyces (Actinomyces odontolyticus, Actinomyces graevenitzii, Actinomyces oris, Actinomyces naeslundii), Neisseria (Neisseria flavescens/perflava, Neisseria mucosa, Neisseria macacae) and Gemella (Gemella sanguinis, Gemella haemolysans) were found in 22%, 14%, 10% and 5% of the isolated bacteria, respectively. The percentages of all other recovered species mainly belonging to the genera Veillonella, Capnocytophaga and Campylobacter, ranged from 2-8%. As depicted in Fig. 5B, the aPDT with VIS + wIRA and H. perforatum -which was rinsed prior to aPDT -reduced the number of viable bacterial species from 31 to 8, comprising S. sanguinis/parasanguinis (31%), S. mitis/oralis (23%), R. mucilaginosa (22%), N. mucosa/macacae (12%) and Granulicatella elegans (12%). In the presence of reduced microbial richness, the species diversity of the surviving bacteria among the probands was also strongly affected. The most prevalent species in the untreated control group were Streptococcus mitis/oralis, R. dentocariosa and Veillonella parvula/dispar (detected in five volunteers), followed by A. odontolyticus and R. aeria (retrieved from four volunteers). All other species were detected in three volunteers or fewer. After pre-treatment with H. perforatum, rinsing and subsequent aPDT with VIS + wIRA, no cultivable bacteria were detected in four of the volunteers. The two remaining positive samples were mainly comprised of S. sanguinis/parasanguinis and G. elegans, whereas seven different species were only detectable in the second volunteer.  Relative distribution among the volunteers and total distribution (in percentages) of bacterial species among six probands after two-hour (2 h) initial adhesion following aPDT using VIS + wIRA and H. perforatum. Graph A depicts the bacterial composition of the untreated negative controls after initial adhesion. Graph B shows the bacterial diversity after aPDT against initially adhered bacteria. Consistent color coding was used for the study participants and bacterial species, as indicated by the schemes at the right of panels, respectively. www.nature.com/scientificreports www.nature.com/scientificreports/ Shift of bacterial community composition after apDt application to the mature oral biofilm. Interestingly, after aPDT with VIS + wIRA in the presence of H. perforatum, the total microbial load of the biofilms was reduced by 100%. The microbial richness after pre-treatment with H. perforatum and aPDT with VIS + wIRA and the bacterial composition of the controls (negative: untreated, positive: treatment with 0.2% CHX) are illustrated in Figs. 8A-C, 6B and 7C-E. Regarding the microbial richness within the mature biofilms, the untreated control, the CHX-treated positive control and the aPDT-treated biofilms comprised 35, 27, and 30 different bacterial species, respectively.
In the positive control (Fig. 8B), the microbial richness was reduced to 27 species after treatment with 0.2% CHX. Although the counts of Eikenella corrodens, S. gordonii, and R. dentocariosa increased after the CHX Figure 6. Heatmap presenting the absolute distribution (in log 10 /cm 2 ) of bacterial species among six study probands after two-hour (2 h) initial adhesion and three-day (3 d) biofilm formation following aPDT using VIS + wIRA and H. perforatum. The photodynamic efficiency against oral microorganisms after initial adhesion and three-day biofilm formation is demonstrated in panels A and B, respectively. The bacterial composition of the following treatment groups is demonstrated for two-hour-and three-day old oral biofilms: untreated negative control, 0.2% CHX-treated positive control, aPDT after rinsing off H. perforatum, aPDT without rinsing off H. perforatum. Proband numbers for each treatment group (n = 6) are shown in columns and variables (bacterial species and genera) are shown in rows. The variations in the colors depicted on the color scale bars on the right indicate how data values change from very low to extremely high. (2019) 9:20325 | https://doi.org/10.1038/s41598-019-56925-7 www.nature.com/scientificreports www.nature.com/scientificreports/ treatment, all of the other bacterial counts within the positive control decreased. A total of ten species could not be detected after treatment with CHX, whereas Abiotropha defectiva and Atopobium parvulum could only be detected after the treatment with CHX.
Regarding the species distribution among the probands, S. mitis/oralis, R. mucillaginosa and V. parvula/dispar were detected in all six volunteers of the untreated control group, with a total distribution ranging from 8 to 11%. Capnocytophaga spp. (7%) was detected in five volunteers, whereas the other bacterial species were recovered from 1-4 probands. After the aPDT treatment, S. mitis/oralis, S. sanguinis/parasanguinis, R. mucillaginosa, V. parvula/dispar and Campylobacter concisus were detected in all six volunteers, with a total distribution ranging from 4 to 11%. All other species appeared in 1-5 volunteers and ranged from 1 to 4%. After aPDT treatment, a total of eight aerobic and three anaerobic microorganisms increased in their relative distribution compared with the CHX treatment, where only three aerobic microorganisms increased. In terms of the new appearance of bacteria after the aPDT with H. perforatum, two aerobic and two anaerobic bacteria emerged compared with the CHX treatment, where only one aerobic and one anaerobic bacteria appeared.

Discussion
The present study has established an effective method for microbial photoinactivation combining VIS + wIRA with the natural photosensitizing agent H. perforatum to treat in situ-formed initial and mature oral biofilms for the first time. The aPDT combining VIS + wIRA and photosensitizing agents such as TB and Ce6 was reported to successfully photoinactivate oral pathogenic microorganisms situated even in the deepest niches of in situ supragingival and subgingival biofilms in earlier own reports 14,15,17,33 . In an in vitro study investigating the effectiveness of H. hypericum-mediated aPDT on E. faecalis culture in planktonic form and biofilm, a significant microbial . Pie charts demonstrating the total distribution (in percentages) of bacterial genera among six probands after two-hour (2 h) initial adhesion and three-day (3 d) biofilm formation following aPDT using VIS + wIRA and H. perforatum. Graph A depicts the bacterial composition of the untreated negative controls after initial adhesion. Graph B shows the bacterial diversity after aPDT against initial biofilms. Graphs C-E show the bacterial composition of three-day old biofilms divided into the following groups: the untreated negative control (C), after treatment with 0.2% CHX (D), and after aPDT (E) with VIS + wIRA and H. perforatum (was rinsed off prior to aPDT). Consistent color coding was used for the bacterial genera, as indicated by the schemes at the right of panels A-E, respectively. www.nature.com/scientificreports www.nature.com/scientificreports/ reduction of up to 5 log counts was detected 34 . The innovative contribution of this study is that, to the best of our knowledge, H. perforatum -widely known for its anti-depressant properties -is here studied for the first time in combination with VIS + wIRA as a light source to treat in situ oral biofilms [35][36][37][38] .
For an effective aPDT light source, visible light (VIS) with an accessory orange filter within the range of 570-780 nm was combined with water-filtered infrared-A (wIRA) to accentuate the amount of infrared-A radiation within the range of 780-1400 nm 39 . Since PS excitation is only possible within the range of VIS, the use of wIRA alone would not have yielded the desired photodynamic outcomes 39 . The advantages of using VIS + wIRA contrary to VIS alone have already been outlined in the literature 9 . The emission of energy in the form of light results in electron transfer reactions between the substrate and the PS, with the latter entering an electronically excited state 40 . The leading antimicrobial effect of aPDT is the subsequent production in elevated rates of reactive oxygen species (ROS) such as superoxide ion (O 2 − ), hydrogen peroxide (H 2 O 2 ) and free hydroxyl radicals ( . HO), all yielded via Type I mechanisms 41 . In a Type II reaction, single oxygen radicals ( 1 O 2 ) are yielded after the reaction of triplet state PS and oxygen (O 2 ) 42 .
Due to the short lifetime of ROS (about 10-320 ns), their diffusion after using conventional light sources (laser and non-laser) is limited and thus the direct aPDT-induced damage remains localized to the PS-infiltrated area 43 . VIS + wIRA also allows for the release of the protective protein ferritin 44 to prevent cell destruction. The high safety of VIS + wIRA radiators can be guaranteed for treatment times of up to 30 min 45 , although 5 minutes proved to be a potent average aPDT duration in our report with a focus on daily clinical use. To facilitate a better distribution of photosensitizer within the bacterial cells and to achieve clinically meaningful treatment times, the biofilms were pre-exposed to H. perforatum for 2 min in the dark. In a systematic review, Fumes et al. described pre-irradiation times for methylene blue (MB) between 1 and 15 minutes, with optimal outcomes between 5 and 15 min 46 . However, there is no published data on the pre-irradiation incubation of H. perforatum, which exhibits enhanced cell penetration when diluted in ethanol 47 . Finally, VIS + wIRA raises levels of physiological activity of the regulatory enzyme cytochrome c oxidase with absorption maxima at 620, 680, 760 and 825 nm 48 . www.nature.com/scientificreports www.nature.com/scientificreports/ Due to its numerous advantageous properties, H. perforatum constitutes a multipurpose plant 49 and is considered one of the most well-known natural photosensitizers in different medical fields 50 . Interestingly, H. perforatum has proven extremely effective in the treatment of non-melanoma skin cancer as well as other dermal and internal cancer species 18,51,52 . Besides its excellent anti-oxidant and anti-inflammatory effects 53 , H. perforatum has been successfully applied in dentistry for the pain management of trigeminal neuralgia 54 . Furthermore, hypericin and hyperforin -the main components of H. perforatum -are responsible for enhanced anti-bacterial activity against Gram-positive and Gram-negative microorganisms in both planktonic and biofilm forms 55 . As a result of its beneficial features, H. perforatum can be integrated in a greater number of clinical photoinactivation protocols compared with Ce6, TB, ICG and MB 18,35,37,56 . When administered systematically, H. perforatum inevitably interacts with several cytochrome enzymes, inducing lower plasma levels of anti-cancer agents or other medicines 57 . Nevertheless, local application of H. perforatum is associated with fewer side effects regarding cytotoxicity and cross-reactivity with other medicaments 58 . Even after the combined oral application of H. perforatum using both a single-and double-dose of UV radiation, visible light, and solar-simulated radiation, no significant pathogenic alterations could be detected at erythema threshold levels 59 . However, a recent report highlighted the increase of round fibroblastic cells, which indicated high cytotoxicity in concentrations of H. perforatum higher than 10-50 mg/ml 60 .
Solvent extraction is the most common means of extracting Hypericum from the flowers of H. perforatum (St John's Wort). Solvents include methanol, ethanol, ethyl acetate and acetone. Scientific studies indicate when using extraction solvents containing more than 50% ethanol or methanol in water the content of the flavonoids in particular Hypericum and Hypericin in the extracts is similar independent of the extraction solvent 61 . H. perforatum extracts are often purified by removing the solvent and then dried into powders prior to commercial sale. This project required a crude liquid extract which contained the solvent for further clinical trials. For this reason methanol, ethyl acetate and acetone were not used as the solvent for the extraction of Hypericum due to the potential toxicity of the solvent in the crude extract. In order to extract bioactive constituents from H. perforatum, organic solvents were preferred, as ethanolic H. perforatum extracts showed the highest anti-bacterial potential against diverse Gram-positive and Gram-negative bacteria 37 . In the literature, other H. perforatum hydroethanolic extracts such as tinctures with a high alcohol content ranging from 45 to 50% have been widely used in complementary medicine 62 . In a recent report, 60% or above of ethanol was utilized to obtain the highest yields of lipophilic hyperforin 63 . However, the use of 100% (absolute) ethanol led to a reduction of specific hydrophilic components such as flavonoids 18 . Since bacteria in biofilms are more resistant to antimicrobials than their planktonic counterparts, we did not expect a significant decrease of biofilm viability after its exposure to 16% alcohol solution for seven minutes as detected by CFU. To confirm our assumption, we ran primary experiments resulting in no significant differences in CFUs between biofilms after 7-min treatment with 16% alcohol and the untreated control ( Supplementary Fig. S1). Hence, a bactericidal effect of alcohol at the tested concentration can be excluded. It is recommended that future studies examine the role of alcohol in detail using live dead staining along with CFU determination. A bactericidal effect of the alcohol on the mature oral biofilm is not possible after an incubation time of seven minutes as detected by the CFU 64 . The authors reported that an alcohol concentration of 30% and incubation time of 8 hours were required to detect bactericidal effects against biofilms formed by Klebsiella pneumoniae and Pseudomonas aeruginosa after 40 hours. It should be taken into consideration that mature multispecies oral biofilms are more resistant to alcohol than mono-species biofilms tested in the aforementioned study. Ultrasonication (70% amplitude) using a digital ultrasonic bath (Sonorex Digital 10 P from Bandelin, Berlin, Germany) was applied to dissolve adherent bacteria from substratum. This ultrasonication power treatment for 4 min on ice was highly effective in killing adherent anaerobic bacteria such as Porphyromonas gingivalis, as reported in a previous own study 65 .
In a recent study, a promoting effect for periodontal healing was found for photosensitizer methylene blue (MB), which was dissolved in 20% ethanol during aPDT 13,47 . Presumably, this can be attributed to the up to 5-fold increase in survival time of ROS 66 . Another study showed that 20-30% ethanol is necessary to prevent the resuscitation of multispecies biofilms 67 . This supportive antimicrobial effect of ethanol can be attributed to the manipulation of the bacterial membrane and the denaturation of proteins 68 . In earlier reports, it has been shown that lipophilic photosensitizers are much more efficient than hydrophilic ones with the same ROS supply 69,70 . In contrast to conventional photosensitizing agents such as TB, ICG, MB and Ce6, the high anti-bacterial efficiency of H. perforatum extract cannot be attributed to mechanisms involving only one specific pure substance. Interestingly, synergetic interactions between three main constituents -namely hypericin, pseudohyperin and hyperforin -are responsible for the high antimicrobial potential of H. perforatum 26,27,71,72 . Nevertheless, their exact mechanism of action remains unknown 26,27 . Further studies are necessary to determine the synergistic potential of specific H. perforatum components.
Obtaining an in situ oral biofilm enabled us to conduct the assays under more realistic experimental conditions, as already described in previous own reports 73,74 . Bovine enamel slabs (BES) represent a favorable habitat for the adhesion of oral microorganisms which is physicochemically similar to human tooth surfaces 75 . Due to substantial differences in biofilm composition and architecture between artificial and natural tooth surfaces, the tests were conducted solely on the latter.
About 700 oral bacterial species are estimated as residing within the oral cavity, of which only 50% have been cultivated thus far 76,77 . Subsequently, the plate count technique as well as the culture media favor the growth of certain types of bacteria 78 . Nevertheless, CFU remains the most widely used gold standard method for the quantification of viable bacteria 79 , which is the focus of this study.
The application of aPDT completely inhibited bacterial growth when the initial and three-day-old biofilms were photodynamically treated in the presence of H. perforatum. This is a very interesting finding, indicating the high bactericidal potential of H. perforatum-mediated aPDT compared with aPDT employing other photosensitizers in previous reports 17 www.nature.com/scientificreports www.nature.com/scientificreports/ polymeric substances (EPS) within the oral biofilm and the chemical composition of the photosensitizers and cell envelope in Gram-positive and Gram-negative bacteria are crucial parameters that can define the degree of photosensitizer diffusion in biofilm biomass. In order to investigate the exact depth of penetration as well as whether non-viable bacteria can survive this procedure, further studies are necessary. Another critical aspect is the possibility that aPDT favors the survival of resistant bacteria. Although there are no relevant studies in the literature, even if this assumption was true, the application of low concentrations of photosensitizer would carry the same risk of creating a resistance-prone microbial community as treatment with low doses of CHX or antibiotics 1,2,81,82 .
In this study, we have shown that treatment with CHX as well as aPDT after rinsing off H. perforatum reduced the number of viable bacterial species, albeit not significantly. Nevertheless, the shift to a predominately aerobic environment after aPDT holds strong importance as it implies the presence of a high alteration potential of H. perforatum-mediated aPDT on biofilm synthesis. Caries and periodontitis are biofilm-associated oral infections that can be controlled after favorably changing the composition of the pathogenic microbial ecosystems into healthier microbial communities 23,77,83,84 . Ascertaining which alterations of the biofilm composition are advantageous for the development of a 'healthy' microbial environment constitutes an alternative treatment approach in case the complete mechanical removal of the biofilm poses clinical challenges and should be the focus of future investigations [85][86][87] .
Interestingly, H. perforatum had the highest efficiency as a photosensitizer when it was not rinsed prior to aPDT. This could be explained by the fact that the prolonged contact of H. perforatum with the oral biofilm could facilitate the maximum penetration of the biofilm through diffusion or endocytosis 4 . Taking the short lifetime of the ROS (about 10-320 ns) into consideration, the time extension of the synergistic interaction between diverse components in H. perforatum total extract and biofilm constituents such as EPS may be crucial for the aPDT-induced antimicrobial efficacy 17,21 . The highest antimicrobial effect can be achieved when the photosensitizer remains as close as possible to the targeted bacteria 43,88 .
To sum up, this study has shown that H. perforatum as a PS in combination with VIS and wIRA has a strong ability to eradicate in situ oral biofilms. The null hypothesis was rejected. The results of the study highlight the already-known positive effects of aPDT in the dental field and introduce H. perforatum as a potent photosensitizer for clinical use.

Selection of study volunteers and test samples. After the review and approval of the study protocol
Radiation source and photosensitizer. The obtained biofilm samples were illuminated by broadband VIS + wIRA radiator (Hydrosun 750 FS, Hydrosun Medizintechnik GmbH, Müllheim, Germany) containing a 7 mm water cuvette 14,17 . For the absorption of the infrared-B and -C wavelengths emitted by the halogen lamp, an accessory water filter (dimensions: length: 28 cm, width: 27 cm, height: 28 cm) using 750 W (voltage: 230 V, 50-60 Hz) was inserted into the light path. The radiator also included an orange filter, namely BTE 31, with a diameter of 10 cm and an efficiency level of 200 mW/cm². The BTE 31 filter enabled more than a www.nature.com/scientificreports www.nature.com/scientificreports/ doubled-weighted efficient integral illumination with reference to the absorption spectrum of protoporphyrin IX. In order to prevent superficial overheating, the absorption bands 944 and 1180 nm were also filtered. The total spectrum of the water-filtered absorption ranged from 570 nm to 1400 nm, with local minima at around 970 nm, 1200 nm and 1430 nm, respectively 90 . The total amount of unweighted (absolute) radiation applied to the biofilms for 5 minutes equaled 200 mW cm −2 VIS + wIRA, which comprised 48 mW cm −2 VIS and 152 mW cm −2 wIRA. This broad-spectrum light source provided optimal absorption for the tested photosensitizer, namely H. perforatum ethanol (15.8%) extract (St. John's Wort, Quality Herbal Extracts, Cooranbong, Australia). Pseudohypericin -the main component of H. perforatum -can be activated within the UVA (and UVA1) visible absorption range, while absorption peaks are mainly exhibited at red wavelengths (570-700 nm) 91 . The visible absorption maximum (λ max ) of H. perforatum is 598 nm 92 . For the assays, H. perforatum extract was diluted in 0.9% saline (NaCl) to a final concentration of 32 mg ml −1 and a final alcohol concentration of 15.8% (v: v). Prior to use, the original extract solution was stored in a light safety glass at 4 °C in the dark for up to fourteen days to block any light-induced photochemical attenuation. extraction of H. perforatum. Wild-growing Hypericum perforatum was harvested during the flowering phase. The herbaceous material was dried, cut, and stored until further processing. The crude extract was produced by percolation using 95% aqueous ethanol as a solvent (menstruum). The milled herbal material was moistened with some of the menstruum and packed firmly in a cylindrical percolator. Additional menstruum was added to the percolator and the herb was allowed to macerate for 24 hours before the percolation was commenced. Percolation comprised the slow passage of menstruum through the column of drug. The maceration and percolation were performed at 30 °C. The percolation was repeated until the material was exhausted. The extractions were combined and the solvent recovered under vacuum evaporation to yield a final ratio of one part drug per one part solvent (1: 1).

Ultra-high-performance liquid chromatography coupled with high resolution mass spectrometry (UpLc-HRMS).
The H. perforatum extract was submitted to UPLC-HRMS/MS analysis on an AQUITY system (Waters) connected to an LTQ-OrbitrapR XL hybrid mass spectrometer (Thermo Scientific) equipped with an electrospray ionization (ESI) source and operated in negative mode. A UPLC separation gradient was developed to efficiently resolve all compounds for a qualitative analysis. The flow rate was set at 0.4 mL/min and the solvent system was (A) water 0.1% formic acid and (B) acetonitrile 0.1% formic acid. The elution program was 2% B for 2 minutes, 100% B for 18 minutes, and held for 2 minutes. After returning to 2% B for 1 minute, column equilibration was performed for 4 minutes at the end of the run. The injection volume was set to 10 μL and samples were injected at 0.2 mg/mL in water-acetonitrile solution (1:1) on a Fortis, C18 (100 × 2.1 mm i.d, 1.7 μm particle size). The HRMS and HRMS/MS data were acquired in negative mode over a 100-1000 m/z range. The MS profile was recorded in full-scan mode (scan time = 1 micro scans and maximum inject time = 500 ms). The ESI conditions were as follows: capillary temperature 320 °C; capillary voltage −40 V; tube lens −120 V; ESI voltage 2.7 kV. Nitrogen was used as sheath gas (40 Au) and auxiliary gas (8 Au). For the HRMS/MS acquisitions, a data-dependent method including the detection (full scan) and fragmentation of the three most intense peaks per scan was used. The mass resolving power was 30,000 for both levels and the normalized collision energy (CID) in the ion trap was set to 35.0% (q = 0.25) for the HRMS/MS experiments. Chromatographic and spectrometric features were used for identifying extracts constituents such as retention time (Rt), polarity, accurate m/z, proposed elemental composition (EC) and ring double bond equivalent (RDBeq) values as well as HRMS/ MS spectra and derived fragmentation motifs. The raw data were acquired and processed using XCalibur 2.2.4 software from Thermo Scientific. apDt protocol for in situ oral biofilms. Each participant twice carried an individual BES-loaded upper jaw acrylic device for the assays involving the initial adhesion (two hours) and biofilm formation (three days). After obtaining the in situ biofilms, the devices were removed from the oral cavity. Subsequently, BES were detached using sterile tweezers and a dental probe to avoid disturbing the intact in situ biofilms. Each of the BES was rinsed for 30 seconds with sterile 0.9% NaCl to ensure the displacement of non-adherent bacteria. One BES served as a positive control and was treated with 0.2% CHX, while another two BES treated with 0.9% NaCl and H. perforatum without aPDT were used as negative controls. One biofilm-covered BES was preincubated in H. perforatum, which was rinsed off for 30 seconds with 1 ml 0.9% NaCl prior to aPDT ex vivo using VIS + wIRA to allow for optimal depth of the irradiation. The remaining BES was also preincubated in H. perforatum, which remained during aPDT ex vivo using VIS + wIRA. The preincubation time for each BES in H. perforatum at a concentration of 32 mg ml −1 was two minutes in multiwell plates (24-well plate, Greiner bio-one GmbH, Frickenhausen, Germany) in the dark as described elsewhere 17 , while the irradiation by VIS + wIRA lasted five minutes at 37 °C (Fig. 2). After aPDT, the BES were inserted into multiwell plates with 1 ml 0.9% NaCl and the quantification and identification of adherent microorganisms were performed by the CFU and with MALDI-TOF, respectively. The activity of the photosensitizer remained stable for six months of storage at low temperature (3 °C) in the dark using a light-tight bottle. The participants carried the BES twice for 72 h and 2 h, respectively.
Quantification of the adherent oral microorganisms within biofilms. The margins and the bottom dentine surfaces of the biofilm-loaded BES were brushed down with small sterile foam pellets (Voco GmbH, Cuxhaven, Germany) to displace non-adherent microorganisms. Subsequently, the BES were washed with 0.9% NaCl for at least 10 seconds to remove the remaining non-attached microorganisms. For the dilution series, each treated BES was transferred into Eppendorf tubes (Eppendorf GmbH, Wesseling-Berzdorf, Germany) containing 1 ml 0.9% NaCl, ultrasonicated for 3 minutes in 1 ml NaCl on ice and finally vortexed for at least 30 seconds to dislodge the biofilm microorganisms from the BES surface. Subsequently, the suspensions untreated BES (negative www.nature.com/scientificreports www.nature.com/scientificreports/ control), CHX-treated BES (positive control) and all aPDT-treated BES were diluted from up to 1:10 3 (initial biofilm) and up to 1:10 6 (mature biofilm) in 0.9% NaCl. From each dilution series of the differently treated BES three different dilutions were plated on separate agar plates. Thereafter, the microorganisms were cultivated and identified as previously described 93 . In brief, for the CFU estimation of the aerobic and facultative anaerobic bacteria Columbia blood agar plates (CBA; Becton, Dickinson, Heidelberg, Germany) were used at 37 °C and 5% CO 2 for 3-5 days. Anaerobic bacteria were cultivated on yeast-cystein blood agar plates (HCB; Becton, Dickinson, Heidelberg, Germany) and incubated for ten days at 37 °C (anaerobic chamber; Genbox bioMérieux SA, Marcy l'Etoile, France). The CFU were finally counted per ml of the original suspension using Gel Doc EQ universal hood (Bio-Rad Life Science Group, Hercules, CA, USA). Each measurement was repeated, and the surviving bacteria were counted by eye if necessary.
Identification of the adherent oral microorganisms within biofilms. The recovered bacterial colonies were sub-cultivated to yield pure cultures. In order to identify certain bacterial species, MALDI-TOF analysis in a MALDI Biotyper Microflex LT (Maldi Biotyper, Bruker Daltonik GmbH, Bremen, Germany) was applied. The Biotyper 3.0 Software obtained mass spectra according to the manufacturer's instructions and compared the acquired spectra with a total of 3,740 spectra of the reference database (representing 319 genera and 1,946 species). The resulting similarities were illustrated by a log score indicating the species level with values ≥ 2.0, whereas values ≥ 1.7 represented identification at the genus level. For values < 1.7, the yielded spectrum exhibited no significant similarity to any database spectrum. The analysis was repeated if the identification of the species could not be confirmed. Moreover, universal bacterial PCR was performed as a supplementary method to identify the surviving oral microorganisms. In the presence of MALDI-TOF scores < 1.7, the following primers were applied: TP16U1: 5′-AGAGTTTGATCMTGGCTCAG-3′ and RT16U6: 5′-ATTGTAGCACGTGTGTNCCCC-3′ followed by sequencing on a 3130 Genetic Analyzer (Applied Biosystems, Life Technologies GmbH, Darmstadt, Germany).

Statistical analysis.
The effects of aPDT using VIS + wIRA and H. perforatum as a photosensitizer were calculated in a detailed evaluation for all six volunteers. For descriptive analysis, mean and standard deviations were computed. For graphical demonstration of the results, boxplots were used. Diagrams that demonstrated the log 10 scale per square centimeter (log 10 /cm 2 ) were selected to visualize the viable bacteria stratified to biofilm age (initial/mature) and type of microorganism (aerobic/anaerobic). A Friedman test was used to depict the differences between the aPDT-treated groups and the controls. Because a nonparametric test lacks sufficient power due to the small number of cases, paired t-tests with Bonferroni correction (multiple testing) were used for pairwise tests of differences among biofilms. Scheffé's method was applied to correct the multiple testing errors through the adjustment of p-values. The statistical analysis was performed using STATA 14.

Data availability
All data generated or analyzed during this study are included in this published article.