Ethics approval was granted by the Ethics Committee of from the Shupyk National Healthcare University of Ukraine (protocol No 8 dated by 7 November 2022) and written informed consent obtained from all participants before inclusion in the study.

All patients were identified as having chronic periodontitis with no systemic disorders after clinical examination which included analysis of the plaque-control record, recording the probing depth and attachment level at six sites around each tooth, and assessment of alveolar bone resorption. No patient had a history of periodontal treatment and none had taken antibiotics in the three months prior to the baseline examination. Sampling of subgingival cytological landscapes was performed by allowing natural biofouling of biologically neutral polyethylene terephthalate (PET) films (2 – 4 x 7 mm) placed in periodontal spaces for five days. Briefly, each film was inserted in a gingival cavity where periodontitis of patients having moderate to advanced periodontitis stage was diagnosed. The patients were instructed to follow their normal oral hygiene but were requested to take care when brushing near the tooth with the film. After five days, the films were removed and placed in a plastic micro-tube and immediately delivered for cytological analysis. When needed, 1 mg/ml DNase (Sigma-Aldrich), dissolved in a buffer containing 20 mM Tris-HCl (pH 7.4) and 10 mM MgCl₂ (Sambrook et al., 1989), was applied for 20 min before staining the samples and subsequent microscopy imaging.

Escherichia coli K12, Klebsiella penumoniae ATCC 10031, Pseudomonas aeruginosa PA01, P. fluorescens SBW25, and Staphylococcus aureus ATCC 29423, were obtained from laboratory stock collections (Moshynets et al., 2020; Chernii et al., 2021; Moshynets et al., 2022b; Volynets et al., 2022). The Ukrainian hospital isolates K. pneumoniae UHI 489 and UHI 117 were from previous work (Moshynets et al., 2022a) and three oral streptococcal opportunistic pathogenic isolates, Streptococcus gordonii UHI1, S. sanguinis UHI1 and S. mitis UHI1, were isolated from samples collected from patients of the Maxillofacial Surgery Department of the Kyiv Regional clinical hospital with purulent-inflammatory processes and identified using a Bacteriological Analyzer Vitek 2 Compact v9 (bioMérieux, France). Bacteria were incubated aerobically at 37°C except for P. fluorescens SBW25 which was cultured at 28°C. Over-night shaken Luria-Bertani (LB) liquid cultures provided 300 µl inocula for biofilm assays which were undertaken in static 30 ml microcosms containing 5 ml media. These were incubated at 28°C or 37°C and biofilms harvested after three days. Solid-liquid interface biofilms formed on the bottom surface of microcosms, except for Pf SBW25 where air-liquid interface biofilms were formed (Moshynets et al., 2022c).

Biofilm and cytological analysis was performed using confocal laser scanning microscopy (CLSM). Biofilm samples obtained from microcosms, cytological samples obtained with a spoon excavator, and subgingival biofilms formed on biologically neutral polyethylene terephthalate (PET) films, were transferred to microscopic glass slides prior to staining. Fluorescent stains were used to visualize intracellular DNA (iDNA) and extracellular DNA (eDNA), amyloids, total protein, and Candida cells. Stain combinations were based on complementary excitation/emission (Ex/Em) wavelengths which allowed individual stains to be visualized using separate CLSM channels.

iDNA was visualized with 5 µM SYBR Green solution (Thermo Fisher Scientific) which can be used to stain live cells with an intact membrane or dead cells or those with a damaged membrane and was used as a proxy for bacterial cells. 2 µg/ml Ethidium bromide solution (Sigma-Aldrich, UK) was also used to visualize iDNA because under the conditions we used, the eDNA signal was found to be very diffused compared to the signals coming from intact cells. eDNA was visualised with 20 µg/ml Propidium iodide in water, or 1 mM of the experimental stain 986 (Moshynets et al., 2022c) (supplied by SM Yarmoluk) in DMSO, which do not penetrate intact cells. Double-staining eDNA with Propidium iodide and SYBR Green is not possible as this dye combination results in fluorescence resonance energy transfer (FRET) reducing SYBR Green fluorescence and increasing Propidium iodide fluorescence (Barbesti et al., 2000). Amyloids were visualised with 1 mM AmyGreen solution [of our own synthesis (Moshynets et al., 2020)] in DMSO and total proteins with 0.5 µg/ml Thiazine Red R (Sigma) in 0.1% acetic acid. Carbohydrates (β(1-3) glucans) and Candida cells (β(1-4) glucan, i.e., chitin) were visualised with 2 µg/ml Calcofluor White (Sigma-Aldrich, UK) in water (Domenech et al., 2013; Rezende et al., 2019).

No additional washing after staining was performed in order to limit the physical disruption of samples during preparation, and a cover slip was placed over the samples before imaging. CLSM analysis was undertaken using a Leica TCS SPE Confocal system with coded DMi8 inverted microscope (Leica, Germany) and Leica Application Suite X (LAS X) v3.4.1. Representative images were acquired using a maximum of three channels for SYBR Green or AmyGreen (Ex/Em 488/490 – 580 nm), Ethidium bromide, 986, Propidium iodide or Thiazine Red R (532/537 – 670 nm), and Calcofluor White (405/450 – 500 nm), with total pixel numbers determined for each stain (channel) using LAS X. Data are shown as total abundance or as the relative abundance compared to iDNA levels.

Analysis of biofilm samples by FISH followed earlier work (Hannig et al., 2007). Oligonucleotide probes for eubacteria, streptococci and the Bacteroides–Porphyromonas–Prevotella (BPP) group were synthesized commercially, and 5’-end labelled with fluorochromes (Syntol, Russia). EUB 338 (5’-GCT GCC TCC CGT AGG AGT-3’) was 5’ labelled with FAM (Ex/Em 488/491 – 575 nm) and used to visualize bacterial eDNA (Amann et al., 1990). STR 405 (5’-TAG CCG TCC CTT TCT GGT-3’) was 5’ labelled with Сy5.5 (Ex/Em: 635/645 – 800 nm) and used to visualize eDNA from oral streptococci (Paster et al., 1998). The Bacto 1080 probe (5’-GCA CTT AAG CCG ACA CCT-3’) was labelled with Acridine (Ex/Em 405/407 – 496 nm) and was used to visualize eDNA derived from the BPP group (Doré et al., 1998). Samples were incubated with 1 mg/ml of lysozyme derived from chicken egg white (47,000 U/mg, Sigma-Aldrich, Germany) in 0.1 M Tris–HCl, 5 mM EDTA (pH 7.2) solution, for 10 min at 37°C, in order to permeabilize cell membranes. For subgingival samples and biofilms formed on PET films, the lysozyme step was omitted to limit probe penetration of cells. These were fixed overnight in 4% paraformaldehyde in phosphate-buffered saline (PBS, 130 mM NaCl, 3 mM KH₂PO₄·2H₂O, 7 mM Na₂HPO₄·12H₂O, pH 7.2) at 4°C and then washed with phosphate-buffered saline and stored at -80°C until hybridization. Samples were dehydrated with a series of ethanol washes containing 50, 80 and 100% (v/v) ethanol, for 3 min each before incubation with oligonucleotides at a concentration of 4 ng/µl each in hybridization buffer (900 mM NaCl, 20 mM Tris–HCl (pH 8), 30% (v/v) formamide and 0.1% (w/v) sodium dodecyl sulphate). Hybridization was conducted in plastic microtubes at 35°C overnight using a Dry Block Thermostat TBD-120 (Biosan SIA, Latvia) and samples were then incubated for 20 min at 37°C in wash buffer containing 20 mM Tris–HCl (pH 8), 64 mM NaCl and 0.1% (w/v) sodium dodecyl sulphate. After washing, samples were analyzed by CLSM as described above.

The Vitacell Clinical Laboratory (Ukraine) was used to identify Actinobacillus actinomycetemcomitans, Porphyromonas gingivalis, Prevotella intermedia, Tannerella forsythus (Bacteroides forsythus), and Treponema denticola in subgingival biofilm samples using a Stomatoflor Real Time kit (DNK-Technologia, Russia) (Volinska and Bondarchuk, 2016) and Real-time CFX96 Touch system (BIO-RAD, USA).

Experimental replicates (n) were used and quantitative results presented using whisker plots with the median and mean indicated or as bar plots with the mean and standard deviation (SD). One-Way ANOVA using (www.originlab.com) where p < 0.05 was considered statistically significant.

Different extracellular polymeric substances (EPS) provide structural integrity to biofilms and a variety of EPS including extracellular DNA (eDNA) contribute to polymicrobial supragingival biofilms (Jakubovics et al., 2021). However, in comparison little is known about the biofilm matrix in the subgingival region during periodontitis. As part of our preliminary investigation, we sampled biofilms from patients with periodontitis with a spoon excavator, stained with different combinations of fluorescent dyes (McLaughlin et al., 2017; Moshynets et al., 2020; Moshynets et al., 2022c), and quantified amyloids, total protein, carbohydrates and Candidia cells, and eDNA relative to intracellular DNA (iDNA) levels ( Figure 1 ). This demonstrates the variation seen in biofilm EPS components and the difficulty in establishing normal compositions which may vary within and between patients, and changes resulting from disease progression and possible changes in community composition. While we note variation in total proteins and carbohydrates and Candidia cells in our biofilm samples, we feel the relative abundance of eDNA and amyloids is more interesting, as eDNA is a common matrix component of biofilms encompassing cells within a physical network of fibers, and amyloids interact with eDNA and surfaces to promote attachment (Gallo et al., 2015; Schwartz et al., 2016; Tursi et al., 2017).

In our samples, the mean relative eDNA levels compared to iDNA was 7.75x, suggesting substantial amounts of eDNA matrix surround bacterial cells in these biofilms, and this is supported by CLSM biofilm images (see later figures) where eDNA and iDNA are spatially differentiated. However, some caution is also required in interpreting these results, as normalizing data across biofilm samples using iDNA levels may not always be appropriate (e.g., mature biofilms may have a higher microbial load compared to the eDNA matrix compared to younger biofilms; higher levels of host immune response might have lysed more cells without altering the matrix, etc.).

Functional amyloids (i.e., amyloid fibers) are produced by a number of periodontal pathogens including Streptococcus mutans and representatives of the red complex of periodontal pathogens (Besingi et al., 2017). One of these, Porphyromonas gingivalis, was shown to utilize amyloids for biofilm formation, with at least six variants of amyloid-like fimbrillin (FimA) being produced by this species (Amano et al., 2004; Cueno et al., 2015). Another member of the red complex, Tannerella forsythia, also produces the leucine-rich repeat protein BspA, which is potentially amylogenic as predicted by the Waltz algorithm, and is an important adhesin for host receptors such as glycoprotein 340 (Loimaranta et al., 2009). In our samples, the mean relative amyloid level was 0.007x suggesting very low levels of these fibers were present in the biofilms. We investigated this further by determining amyloid levels in single-species in vitro biofilm produced by Klebsiella pneumoniae, Pseudomonas aeruginosa, Staphylococcus aureus, and three oral-associated streptococci including Streptococcus gordonii, S. sanguinis and S. mitis, recognized as the early tooth surface colonizers (Rickard et al., 2003), with Escherichia coli and an environmental isolate of Pseudomonas fluorescens included as additional biofilm reference strains ( Figure 2 ).

Amyloid levels relative to iDNA in the single-species biofilms varied from 0.04x for the weak producer Kp UHI 486 we have characterized before (Moshynets et al., 2022a) up to 0.71x and 0.74x for amyloid-superproducers Ec K12 and Sm UHI1. Sa ATCC 29423, Sg UHI1 and Ss UHI, also known to produce considerable amounts of amyloids and are early colonizers of teeth (Ximénez-Fyvie et al., 2000; Larsen and Fiehn, 2017), had similarly high relative levels of amyloids in these single-species biofilms. Interestingly, we note that the two pathogenic K. pneumoniae isolates tested here produced lower levels of amyloids compared to the ATCC reference strain that may now be laboratory-adapted.

The relatively low level of amyloids in our subgingival biofilms might be explained by a low number of amyloid producers or by downregulation of amyloid genesis to avoid or mitigate the immune response, as described for BspA in the T. forsythia infection model (Inagaki et al., 2005). Further evidence of an immunogenic effect of amyloid fibers in biofilms has been seen when bound to eDNA (Gallo et al., 2015; Tursi et al., 2017). However, we found lower eDNA levels relative to iDNA in our mature single-species biofilms of 1.41 – 3.76x compared to the subgingival biofilm samples we had sampled earlier with eDNA levels of 7.84x ( Figure 3 ). Interestingly, amyloids are known to stabilize eDNA in biofilms and, reciprocally, eDNA promotes amyloid expression in Enterobacteriaceae (Schwartz et al., 2016; Taglialegna et al., 2016; Yan et al., 2020). However, this may not occur or be as relevant in subgingival biofilms as we have observed high levels of eDNA and low levels of amyloid. This in turns raises the question of whether all of the eDNA we have identified is of bacterial origin or if some of it is derived from host cells.

One of the leading eDNA producers is P. aeruginosa which utilizes eDNA for the initiation and maturation of biofilms (Subendu, 2020; Campoccia et al., 2021; Moshynets et al., 2022b) and is also capable of recruiting host eDNA during maturation (Walker et al., 2005). As our subgingival biofilms contained even more eDNA than that seen in our Pa PA01 biofilms or biofilms produced by other reference strains, it is likely that subgingival biofilm development also incorporates host eDNA during development. eDNA is a common component of natural dental plaque formation (Rostami et al., 2017) and is probably also common in the subgingival niche where the most obvious source are neutrophils attracted by an inflammasome during periodontal diseases. A number of studies have described the role of extracellular DNA traps (NETs) produced by neutrophils during periodontitis (Vitkov et al., 2009; Yousefi et al., 2020). However, the contribution of host eDNA derived from NETs to the development and persistence of pathogenic biofilms in periodontitis remains unclear.

The importance of local environmental conditions and the econiche for biofilm development and community function has been recognized since the pioneering work in microbial ecology by Winogradsky (Moshynets et al., 2013). However, the modern understanding of biofilm structure, development, and function, has emerged largely from investigations of in vitro model systems using a variety of stationary microcosms (Koza et al., 2011; Robertson et al., 2013; Koza et al., 2017) or dynamic flow chambers (Chin et al., 2006; Tolker-Nielsen and Sternberg, 2011). Naturally occurring biofilms develop during colonization of an appropriate surface by multiple species assembling from the local environment and selected by niche physical-chemistry and topography, available resources, competition and predation. This process can be investigated by direct sampling and might reveal important interactions not seen when investigating single-species biofilms using in vitro systems, and was first attempted by Cholodny in the 1930s using a glass-slide sampling procedure for soil and subsequently modified by us to use PET films more suitable for modern microscopy techniques (Moshynets et al., 2011; Moshynets et al., 2012). Similar to soil biofilms, the architecture and spatial arrangement of pathogenic biofilms is also an important factor in understanding their persistence and impact on the host. Dental-associated biofilms form on tooth and soft-tissue surfaces following bacterial colonization and can result in complex heterogenic multispecies communities (Rickard et al., 2003). Polytetrafluoroethylene (PTFE, Teflon) carriers have been previously used to sample subgingival bacterial plaque (Wecke et al., 2000).

Dental-associated biofilms can be readily sampled though the procedures used may damage the structure and underlying relationships between host and microbial cells, matrix, and substratum (Wood et al., 2000), and there is little information on pathogenic biofilm structure in situ largely because of this issue. The relationship between the biofilm and host cells and substratum is of particular interest since any effective treatment should be developed based on a realistic understanding of the cytological landscape. Although microbial diversity has been reported for subgingival biofilms, limited information is available for spatial structure (Rickard et al., 2003; Ansiliero et al., 2021) and analysis of crevicular exudate from patients with periodontitis has confirmed the importance of neutrophil extracellular traps (NET)-derived eDNA in pathological processes (Vitkov et al., 2009).

We expected to observe a standard cytological picture of inflammation when sampling periodontal subgingival biofilms from patients using a spoon excavator. However, our CLSM images suggest that these were dominated by bacterial cells and relatively few leukocyte-like cells ( Figure 4 ) and suggests that we sampled the top surface of the biofilm rather than the substratum where more leukocytes might be expected to be found. Visualisation of eDNA and iDNA suggested a degree of heterogeneity in the samples with both dense and relatively sparse regions of cells and/or eDNA evident. We also noted the presence of Candidia cells stained with Calcofluor White ( Figure 4A ) and that relatively few metabolically-active bacterial cells were observed in these subgingival samples ( Figure 4B ).

In order to image undamaged subgingival biofilms, we placed PET films in dental pockets of patients and left them there for five days before recovery ( Figure 5 ). This approach is similar to the use of PFTE or enamel fragments to sample dental plaque (Wecke et al., 2000; Wood et al., 2000) but our films are easier to prepare as it does not require a rigid mount and has no biorisks associated with the use of donated material. The PET film will be rapidly coated in a pre-conditioning layer of host and bacterial material, and it also provides a surface for migratory leucocytes which transform into terminal-effector cells such as neutrophils in infection sites. Once colonized the PET film develops a 3D micro-landscape structure representative of host-biofilm interactions in the subgingival niche during periodontitis. As expected, the micro-landscape was dependent upon inflammation intensity and depth of observation ( Figure 6 ). More bacteria and a broader morpho-diversity and fewer leukocytes were observed when inflammation was moderate and fewer bacteria and a lot of fibrin-embedded eDNA were observed when the inflammation was strong as was observed earlier (Krautgartner et al., 2010). Candida cells dominated the landscape when the inflammation intensity was high and were generally found at superficial locations ( Figure 6D ).

To evaluate the contribution of eDNA to the pathological micro-landscape, PET films were applied to four dental pockets of three patients with initial periodontitis. After five days the films were recovered and cut in half, with one piece treated with DNAse for 30 minutes before staining for total DNA and Candida cells and visualization by CLSM. Representative images from one patient clearly show a substantial reduction of the total DNA signal following DNAse treatment ( Figure 7 ) suggesting that most of the signal was exposed eDNA. It is also notable that far fewer Candida cells remained after treatment, suggesting that eDNA helped retain these yeast close to the surface of the tooth. DNA levels were quantified for each of the four PET samples ( Figure 8 ) and suggests that eDNA contributes up to ~70% of the DNA visualised using this technique. As the total DNA levels determined for the four micro-landscapes visualised here on PET films were not significantly different, we suggest that they are representative of subgingival multispecies biofilms developing normally in patient’s teeth.

A new, experimental, super-heavy DNA-specific stain called “986” with low penetration through intact and damaged membranes specifically developed to visualize eDNA (Moshynets et al., 2022c) was tested using a PET film subgingival micro-landscape ( Figure 9 ). eDNA was mainly associated with the biofilm matrix and a superficial surface of leukocytes, and implies that any leukocyte-associated eDNA released at these sites contributes to the bacterial eDNA and the developing biofilm. Previous work using in vitro biofilms has shown that neutrophils were able to settle onto P. aeruginosa biofilms where they became immobilized, degranulated and transformed their membranes following bacteria direction (Jesaitis et al., 2003). Similar effects where neutrophils activated their functional phagocytosis when attacking S. aureus and S. epidermidis biofilms have also been observed (Günther et al., 2009; Guenther et al., 2009). Neutrophils sensibilization toward bacterial antigens leads to elevated production of reactive oxygen species, antimicrobial peptides and lactoferrin (Arslan et al., 2009; Stroh et al., 2011; Sol et al., 2013), and when associated with periodontitis results in elevated expression of a Cluster of differentiation (CD) markers panel which leads to a neutrophil phenotype change from parainflammatory in the healthy state to proinflammatory in the diseased patient (Oveisi et al., 2019). This results in phagocytosis, degranulation and neutrophil extracellular traps (NET) activation (NETosis) as a strategy to combat or control pathogenic biofilm development. Such a super-activation of the neutrophils can contribute to the pathogenesis of periodontitis (Kornman, 2008) and might potentially enhance biofilm formation and biofilm-associated phenotypic resistance by periodontal pathogens as seen in vitro for P. aeruginosa (Yousefi et al., 2020).

The mechanisms of host eDNA elaboration during NET activation has been recently reviewed (Vitkov et al., 2021). The main NETos activators were bacterial outer membrane vesicles heavily loaded with bacterial LPS (Vanaja et al., 2016) and gingipain cysteine proteases produced by periodontal pathogens including P. gingivalis (Bryzek et al., 2019) [blood monocytes are also capable of extracellular trap release in response to different stimuli (Granger et al., 2017)]. NETs formed during periodontitis lack bactericidal activity due to the proteolytic activity of the gingipains and do not control pathogenic biofilm development. However, periodontitis-associated NETs are able to stimulate periodontal pathogen growth and possibly biofilm development (Bryzek et al., 2019).

In order to evaluate the contribution of host and bacteria-derived eDNA, we undertook fluorescent in situ hybridization (FISH) of PET films using the EUB 338 bacteria-specific probe ( Figure 10 ). Analysis of three samples of subgingival micro-landscapes from two patients shows that bacterial eDNA may contribute 8 – 21% to the total eDNA with the majority derived from host cells activated following periodontitis ( Figure 11 ). It is therefore possible that the release of host DNA is responsible for the consistent persistence and pathology chronization.

We analyzed a randomly-chosen clinical case of periodontitis (patient 5) to determine the origin of bacterial eDNA involved in subgingival biofilm formation by real time PCR and CLSM. The PCR analysis confirmed the presence of pathologically significant periodontal pathogens including Actinobacillus actinomycetemcomitans, Porphyromonas gingivalis, Tannerella forsythensis, Treponema denticola and Candida albicans. FISH analysis of PET films using the Bacto 1080 probe for the Bacteroides–Porphyromonas–Prevotella (BPP) group and the STR 405 probe for the oral streptococci group suggest that these might not be dominant in the subgingival micro-landscape, and as before host eDNA was more abundant than bacterial eDNA ( Figure 12 ). We noticed a difference in the subgingival micro-landscape close to a fibrin clot which are commonly found during periodontitis ( Figure 13 ). No streptococci were present near the clot although an increased number of anaerobic bacteria were found instead. This might be explained by the antibacterial effects of host eDNA and associated proteins which contribute to the clot when the fibrinogen polymerized. Representatives of the BPP group can neutralize the antibacterial activity of NET-associated host defense proteins using gingipain cysteine proteases (Bryzek et al., 2019) which cannot be done by other colonizers or by streptococci in particular. Analysis of the probe signals allowed us to compare the contribution of eDNA derived from both the BPP group and the oral streptococci group with the total bacterial eDNA which suggest that the anaerobes and streptococci may be the main biofilm-producers in the subgingival niche during periodontitis ( Figure 14 ).