The oral environment consists of a multitude of bacterial species living on both soft and hard tissues in complex multispecies communities known as biofilms [1]. Dental plaque, a type of biofilm which forms on the surface of teeth, has been extensively studied for decades due to its association with periodontitis, a chronic inflammatory disease of the tooth-supporting tissues [2]. Periodontitis causes connective tissue attachment loss, inflammation and, if left untreated, tooth loss [2, 3]. One of the factors leading to periodontitis is the accumulation of dental plaque and emergence of dysbiosis therein [2, 3].

Health-related supra- and sub-gingival bacterial communities are often composed of Gramme-positive and Gram-negative early colonisers including Streptococcus, Neisseria, Prevotella, Haemophilus, and Rothia genera, among other less abundant genera, creating a distinct signature of a healthy, symbiotic state [4–7]. However, studies analysing diseased periodontal sites exhibited a shift toward communities containing anaerobic, Gram-negative periodontal pathogens belonging to Socransky's “red complex” and leading to dysbiosis with an associated destructive immune-inflammatory response resulting in tissue damage [8–12].

Fusobacterium nucleatum, an anaerobic commensal member of dental biofilms [13], is present in low numbers in healthy subgingival dental biofilm [4] but is enriched in periodontal pockets [14]. It has been identified as a key bridging organism between the early colonisers and periodontal pathogens [15]. It is also considered an opportunistic pathogen due to its association with systemic diseases, such as cardiovascular disease, ulcerative colitis and colorectal cancer [13, 16], as well as with extra-oral infections which can lead to adverse pregnancy outcomes [13]. Five subspecies of F. nucleatum have been identified to date: animalis, fusiforme, nucleatum, polymorphum, and vincentii [13]. Some authors only recognise four subspecies, classifying fusiforme and vincentii as one subspecies vincentii [17].

Despite the close genomic relatedness of all subspecies, differences in pathogenicity have been recognised, such as host immune response modulation: subspecies nucleatum (ATCC 25586), polymorphum (ATCC 10953) and vincentii (ATCC 49256) were shown to prevent superoxide generation induced by N-formylmethionyl-leucyl-phenylalanine in neutrophil-like HL-60 cells. Additionally, subspecies polymorphum increased the rate of apoptosis in HL-60 cells when compared to the other subspecies [18]. F. nucleatum subspecies were also studied in in vitro multi-species biofilm models. Differences in the incorporation of subspecies into the biofilm were described, where bacterial numbers of F. nucleatum ssp. vincentii and ssp. polymorphum were significantly lower in comparison with ssp. nucleatum [19].

F. nucleatum adhesion proteins have been identified, which mediate coaggregation and biofilm formation with various oral microbes as well as salivary proteins: outer membrane protein RadD with a putative accessory protein Aid1 [20], autotransporter Fap2 [20], porin FomA [20], outer membrane protein CmpA [20], and adhesin FadA [21]. Homologues of adhesins from other bacterial species have also been identified in F. nucleatum in silico, such as YadA-like adhesin originally found in Yersinia species [22].

Surprisingly, despite its structural importance in dental plaque and its role in pathogenic biofilms in colorectal cancer and other diseases, biofilm formation by single F. nucleatum subspecies is poorly understood. Moreover, few authors have used type strains as reference strains to the clinical isolates in studies of F. nucleatum, and many have not reported the subspecies used, making a comparison of these results difficult. While it is clinically relevant to study F. nucleatum in multi-species biofilms, information obtained from single-subspecies F. nucleatum biofilms without the presence of additional binding partners is necessary for better understanding of biofilm-related immunogenic and pathogenic properties of F. nucleatum subspecies and virulence factor expression.

Therefore, this study aimed to investigate single-subspecies biofilm formation abilities by F. nucleatum subspecies along with analysis of differences in adhesins known to mediate biofilm formation and aggregation. A further aim was to utilise a simplified in vitro single-subspecies biofilm model which could be repeated by other researchers, using widely available substrates and substrate coatings, to better understand biological properties of each subspecies. We hypothesised that there is no difference in biofilm formation and architecture among F. nucleatum subspecies.

F. nucleatum as a commensal microorganism, opportunistic pathogen, and emerging oncobacterium in the oral cavity and at extraoral sites has received considerable attention in recent years [16]. Individual subspecies have been differentially related to oral health and disease: FNN and FNA have been associated with disease [34–36]; while FNF and FNV have been identified in healthy sites [13, 34]; interestingly, FNP is found in both healthy [34] and diseased periodontal tissues [35].

Differential involvement of subspecies in health and disease might suggest differences in their virulence properties. It is increasingly evident from in vitro studies that differences in pathogenicity among F. nucleatum subspecies exist, such as differences in coaggregation with other oral bacteria and biofilm formation in multispecies biofilm models [19]. To the authors' knowledge, formation of single-subspecies biofilms by F. nucleatum ATCC type strains has not been addressed in the literature to date, and data presented in this study showed for the first time that not all F. nucleatum subspecies have the ability to form stable single-subspecies biofilms. In our study, FNP did not form biofilms but only small bacterial aggregates. In support of this finding, similar results were obtained by Karched et al., who showed poor single-subspecies biofilm formation by FNP, whilst a strong autoaggregating ability was shown in this subspecies [37].

Biofilm stability of F. nucleatum subspecies has not been addressed in the literature to date. The present study showed considerable biomass variation likely caused by unstable biofilm formation. Under physiological conditions in the oral cavity, F. nucleatum grows as a component of complex multi-species biofilms in the presence of multiple binding partners, especially early colonisers [38]. Periasamy et al. showed that subspecies polymorphum formed stable biofilms with the early coloniser Actinomyces naeslundii, however, it did not form single-subspecies biofilms in saliva [39]. This further supports our observation that FNP does not form stable biofilms on its own. Another factor, which was not considered in the present study, is pH: its manipulation and increase to 8.2 was found to induce biofilm growth in subspecies polymorphum [40, 41]. Investigating the impact of pH on biofilm formation by all subspecies may be an important parameter to include in future experiments.

In our study, formation of stable biofilms was also substrate-specific. First, surfaces were coated with AS, which is used to mimic in vivo conditions and improve adhesion of bacteria to substrates [23]. The advantage of using AS is the elimination of potentially confounding effects of human salivary components and their natural fluctuations, such as antimicrobial peptides, presence of other bacterial species and their products, and extrinsic or systemic factors [42]. AS significantly supported biofilm formation only in case of FNN25 grown on glass and plastic and FNV grown on plastic surfaces. To the best of the authors' knowledge, F. nucleatum single-species biofilm formation on AS coated surfaces has not been reported in the literature to date, however Tavares et al. showed stable single-species biofilm formation by FNF on plastic surfaces coated with human saliva [43]. Using human saliva in future studies of F. nucleatum single-species can help to understand whether and how specific salivary properties may influence the selection and growth of certain subspecies in the oral cavity.

Secondly, surfaces coated with fibronectin, a glycoprotein present in physiological fluids such as plasma and saliva [44], were covered with an intermediate amount of biofilm biomass by all subspecies but FNP. In a previous study, F. nucleatum was shown to adhere to fibronectin [45] and more specifically, subspecies polymorphum ATCC 10953 was found to adhere to fibronectin-coated gingival epithelial cells and fibronectin-coated coverslips as single cells, whilst biofilm formation was not investigated. This supports our finding that single FNP cells adhere to fibronectin coated surfaces, however, studies analysing FNP biofilm formation on fibronectin coated surfaces are missing.

Thirdly, gelatin, a derivative of collagen found in the extracellular matrix of tissues, is a commonly used polymer for coating tissue culture vessels [26]. Benn et al. reported that adhesion of Escherichia coli (E. coli) was improved by gelatin, but was dependent on specific buffering conditions and bacterial strains [46]. In our study, however, gelatin supported only a low amount of biofilm formation and may therefore not be a recommendable coating agent for F. nucleatum biofilms. Additionally, surfaces coated with poly-L-lysine, a cationic coating agent promoting bacterial adhesion by electrostatic interactions as mostly shown in E. coli studies [46, 47], led to higher amounts of biofilm in the present study seen microscopically (Figures 2, 3), but this was not statistically significant in CV assays (Figure 1). Nevertheless, we recommend poly-L-lysine as a biofilm-supporting coating agent for F. nucleatum.

Apart from surface coatings, surface roughness was also found to promote biofilm formation on multiple types of surfaces by providing a larger surface area for adhesion of bacteria and also protecting adherent bacteria from detachment by shear forces [48]. In this study, the highest mean amount of biomass on glass surfaces was measured on sandblasted glass coverslips. Sandblasting, however, makes glass opaque, rendering biofilm analysis by conventional CLSM challenging if an inverted CLSM is not available. Another material frequently used for oral biofilm assays is hydroxyapatite (HA) [49], mimicking dental enamel surfaces. In vivo, periodontal biofilms predominantly adhere to enamel, dentine [50] and cementum [51]. In future studies, utilising dental tissue substrates such as dentine slices or other relevant surfaces such as titanium used in dental implant manufacturing, in combination with inverted CLSM [52] or atomic force microscopy [53, 54] could provide more detailed information on bacterial adhesion and biofilm architecture.

Assessment of the architecture of single-subspecies biofilms showed observable differences among subspecies. Similar observations were made in a previous study investigating Staphylococcus aureus and Pseudomonas aeruginosa, showing strain-specific differences such as large mushroom-like structures, aggregates and differences in thickness [55]. Possible underlying reasons for differences in architecture might be differences in utilisation of available nutrients by subspecies affecting biofilm maturation or subspecies-specific combination of outer membrane proteins influencing coaggregation.

Of note, biofilm structure also differed between microscopy techniques in our study. For example, FNA biofilms analysed with CLSM appeared as mono-layered structures, while FNA biofilms imaged with SEM were shown to be multi-layered. This may have been due to the fixation procedures affecting the degree of biofilm detachment. Fixation preserves cell morphology and the native structure of biofilms [56], however the degree of paraformaldehyde fixation might have been insufficient in our study due to a short fixation period. Longer incubation in paraformaldehyde [52, 56] and comparison with other types of fixatives in future studies may resolve this issue.

In general, biofilm sample preparation for imaging is often a destructive process, involving multiple washing steps, which may affect biofilm structure and introduce artefacts [57]. Presumed water channels, which were observed in the studied biofilms, could also be considered as such artefacts. Additionally, ECM was not observed in any of the samples assessed in this study. Lack of ECM is likely attributed to the preparation technique for microscopy, as it was demonstrated that biofilm fixation and dehydration for SEM can damage ECM [58]. Presence of ECM in F. nucleatum biofilms has been observed in a number of studies, suggesting that F. nucleatum forms biofilms according to the definition by Costerton et al. [59]. Ali Mohammed et al. characterised overall composition of ECM in single-species biofilms formed by F. nucleatum ssp. nucleatum (ATCC 25586) [60]. Subsequently, ECM proteins of the same subspecies were further analysed [61]. Tavares et al. reported ECM surrounding F. nucleatum ssp. fusiforme cells in single-species biofilms [43]. However, the ECM of the remaining F. nucleatum subspecies has not been studied in a single-species biofilm context.

In order to address the limited ability of FNP to form stable biofilms, bioinformatic analysis of known adhesion proteins was performed and showed that a number of orthologues found in this subspecies have very low identity when compared to FNN25, a biofilm forming strain often used as a control in biofilm models [19]. The analysed adhesion proteins RadD, Aid1 [20, 62], Fap2 [63], FadA [21], and CmpA [64] were mainly reported to adhere to other oral pathogens, however they have not been studied in relation to autoaggregation and biofilm formation within the same F. nucleatum subspecies. FomA binds to human protein statherin found in saliva [65]. The YadA-like protein, which was originally reported as an adhesin and virulence factor in Yersinia species and was identified in F. nucleatum [66], adheres to fibronectin and collagen present in the salivary pellicle and extracellular matrix of human cells [22].

Based on our analysis, all proteins were found in the FNP genome, however four of these (CmpA, Fap2, FomA, and RadD) were not well-conserved. Only two proteins, CmpA and Fap2 were detected in all subspecies and were found to share a common autotransporter domain in FNN25. In addition, our in silico analysis found the autotransporter domain in Fap2 in all subspecies. Hence, one may speculate that these adhesion proteins and perhaps the autotransporter domain are important for cell-cell adhesion in single-subspecies biofilms. CmpA and Fap2 had a very low identity in FNP, and this subspecies was the most distant in the phylogenetic tree. This might indicate an inability to form stable single-subspecies biofilms. It is, however, important to note that these bioinformatic results have to be considered with caution and no final conclusions can be drawn due to the small scale of the analysis. Only a selected set of adhesion proteins reported in the literature was analysed and it is likely that other putative adhesion proteins involved in biofilm formation remain to be discovered.

With the method employed here, no autotransporter domains for CmpA were detected in subspecies FNA and FNN23. Additionally, despite a high identity of the YadA-like protein in FNP, a general lack of enhanced adhesion of FNP to fibronectin and gelatin in our study might suggest that YadA in FNP does not play a key role in adhesion to these proteins. Mutagenesis studies could be performed in the genetically modifiable strain FNN23 in order to validate these findings. Mutants lacking each of the above-mentioned adhesion proteins could be studied regarding their adhesion, autoaggregation and biofilm forming ability.

Many authors have utilised planktonic cultures in pathogenicity studies of F. nucleatum, however this might not reflect true virulence of this anaerobic microorganism, as virulence genes have been shown to be upregulated in biofilms [67, 68]. Availability of F. nucleatum single-subspecies biofilm characterisation data might help researchers to perform more biologically relevant F. nucleatum pathogenicity studies using single-subspecies biofilms. Our experiments showed that in general both glass and plastic surfaces supported biofilm growth and thus can be utilised for studies of F. nucleatum subspecies biofilms when transparent substrates are required. Additionally, surface coatings do not seem to be necessary when quantity of the biofilm is not a priority.

Whilst we appreciate that behaviour of laboratory type strains might differ significantly from clinical isolates obtained from patients, we believe that it is important to first characterise virulence properties of subspecies of F. nucleatum, represented by widely available ATCC type strains, to build a knowledge base, which can later be utilised to study virulence of specific clinical isolates. As biofilm formation is one of the virulence properties of bacteria, understanding the differences in pathogenicity of individual F. nucleatum subspecies may reveal new virulence factors as novel therapeutic targets for prevention and treatment of F. nucleatum-mediated infections and diseases.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

MM carried out experiments and data analysis and wrote the manuscript. DB performed bioinformatic analysis and wrote the corresponding parts of the manuscript. MG, SK, IC, and JH were involved in planning experiments, supervised the project, and contributed to the interpretation of the results. JH provided support with statistical analysis and figure preparation. All authors provided critical feedback on the manuscript. All authors contributed to the article and approved the submitted version.

This study was funded by a grant provided by Birmingham Community Healthcare NHS Foundation Trust. Publication of this study was possible thanks to funds provided by the University of Birmingham, Birmingham, UK.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.