$$\rightleftharpoonup{xx}$$
$$\longleftharp{xx}$$,
$$\longrightharp{xx}$$,
Here, methodologies are described to produce a complex multi-species biofilm model representative of gingivitis for co-culturing with HOE tissue to assess the host response following microbial stimulation. This protocol can be adapted for use in investigating host-pathogen interactions between planktonic cells, single-, dual- or mixed-species biofilms, or biofilm-dispersed cells with epithelial tissue from different ecological niches in the human body. The use of tissue models for investigating host-pathogen interactions provides an important advancement to previous co-culture systems that are restricted to 2D monolayers that don't always truly recapitulate the in vivo situation, whereby epithelial tissue contains multiple cell layers16,17,18. Furthermore, challenges associated with in vitro growth of multi-layered tissue models are plentiful, with the use of multiple cell lines and/or growth of cell layers at an air-liquid interface prone to contamination. The commercially available tissues provide a reproducible platform with consistently high-quality models that are comparable between replicates and batches, as shown in previous publications3,7.
For the "preparation of microbial communities for co-culture" section, the inclusion of these microorganisms for the 7-species gingivitis model were chosen based on commonly identified commensals and pathogens associated with the shift from oral health to disease. As with all complex biofilm model systems in vitro, the inclusion of microorganisms is directed by microbiome studies relating to the particular healthy or diseased ecological niche. To this end, we and others have reported the use of an oral health-associated biofilm model containing commensal microorganisms (e.g., Streptococcus and Rothia spp.)3,40,41. Moreover, additional pathogens can be added to these models to create biofilms associated with other diseases. For example, three microorganisms can be added to the 7-species described here to create a disease model associated with periodontitis3, whilst the fungal pathogen Candida albicans can be added to increase polymicrobiality as well as adding a layer of interkingdom complexity33. Indeed, promoting fungal-bacterial interactions can have huge implications on various experimental outputs using such biofilm models in vitro when compared to bacterial-only biofilms42. Ultimately, it is highly recommended that careful consideration be taken when creating a new multi-species biofilm model for such studies. Whilst the included microorganisms should be easily identifiable from extensive literature searches, for most niches and/or diseases, these may not integrate well into a complex model in vitro for different reasons. The following depicts some other potential considerations that should be made:
Biofilm formation dynamics
Biofilm formation in vivo often involves early, intermediate, and late colonization by particular microorganisms. Supra- and sub-gingival dental plaque formation is heavily characterized by initial attachment of the salivary pellicle found on enamel by pioneering species (e.g., Streptococcus species), with intermediate and later pathogens requiring these as a scaffold for colonization43. A similar phenomenon has been proposed in the vaginal environment during bacterial vaginosis, whereby Gardnerella vaginalis is believed to be the initial colonizer, followed by subsequent anaerobes44. However, it is important to note that this may not be the case for other diseases in other ecological niches.
Seeding density for each microorganism
Microorganisms will have different sizes and/or growth dynamics when cultured in vitro; therefore, it is important to consider this during the standardization process. For example, C. albicans is 100-150 times the size of bacterial cells45,46, therefore, it may warrant addition at a lower concentration, e.g., 1 x 106 cells/mL, to such biofilm models.
Species antagonism
Some microorganisms (including the same species but different isolates, laboratory and/or clinical strains) utilized for these models may compete with others during biofilm formation, leading to inhibition of the growth of some microbial species47. A study by Sadiq et al. highlighted how different combinations of microorganisms can influence biofilm biomass resulting from microbial synergy (or antagonism)48. Although investigating such interactions within a biofilm model may be of interest to research groups (e.g., testing pre- or probiotic treatment), others may not account for such antagonism, which could impact multi-species complexity when creating the model.
Duration of biofilm maturation
It is critical to optimize biofilm maturation timeframes as this can depend on the growth dynamics of the microorganisms included and the model system (including substrata) used for culture. Longer maturation times could result in better colonization for later pathogens but more cell death within the models, particularly of the earlier colonizers. Conversely, shorter incubation times could be important if wanting to investigate the effects of immature biofilm models on the host. A study by Brown et al. described how the same 10-species wound biofilm model had different inflammatory profiles in human THP-1 cells when matured for 24 h, 48 h, and 72 h, suggesting the less mature the biofilm, the more pro-inflammatory it is7. Similar results could be observed in multi-layered tissue models. It is also important to note that regular daily media changes are a necessity to minimize cell death in the biofilm, although this can be amended depending on ongoing treatment regimens, e.g., if assessing the effects of prolonged antimicrobial interventions.
For the "organotypic tissue handling, experimental setup, and tissue processing" section, incubation timeframes can be adjusted according to the researcher's needs. Host-pathogen interactions can be investigated at earlier timepoints, e.g., 1-12 h, to later timepoints of 48 h and 72 h, depending on the research question. Secondly, all maintenance media containing antibiotics is supplied; thus, requests need to be made to the company upon ordering to remove these depending on the experimental design. Although the media underneath the insert does not come in direct contact with the upper periphery of the tissue, unless a wound is inflicted in the model6,8, removal of antibiotics may merit consideration if investigating microbial invasion into the tissue.
The microbial material used for co-culture stimulation can also be changed, as discussed above, with the scope to assess planktonic, spent biofilm supernatants (filtered and un-filtered), whole biofilms, or biofilm sonicate incubations with the tissue. For example, previous evidence has shown that tissue models such as those supplied by EpiSkin provide useful models for investigating planktonic fungal-host interactions: C. albicans, Candida auris, Malassezia furfur, and Trichophyton rubrum cells have been shown to attach and interact with peripheral tissue layers in HOE, Reconstructed Human Epidermis (RHE) or Human Vaginal Epithelium, with some of these studies showing stimulation of a host response following fungal infection1,5,6,11,12,49. Similar publications exist for bacterial-host interactions, e.g., biofilm formation of Cutibacterium acnes, a common pathogen associated with the scalp microbiota in dandruff, has been studied on the surface of RHE, when cultured alone and with the fungal skin organism, Malassezia restricta10. Others have investigated the ability of Staphylococcus spp. to attach to RHE, measuring the physicochemical and microbiological characteristics of this bacterial-host interaction50. N'Diaye and the co-authors explored how the human-derived neuropeptide, Calcitonin Gene-Related Peptide, influenced Staphylococcus aureus virulence in the RHE tissue model51. Others have investigated the protective effects of probiotic interventions on Pseudomonas aeruginosa infection of Human Corneal Epithelium14. Meanwhile, from a polymicrobial perspective, different groups have studied the effects of mixed-species biofilms on different tissue substrates2,3,7,13.
Overall, these protocols and studies referenced above document the vast array of applications for organotypic tissue models to investigate host-pathogen interactions. Although studies have shown that EpiSkin models, particularly the RHE skin model, have good applicability for testing cosmetic products for corrosion/irritation52,53,54,55, some limitations exist between these and real-world ex vivo tissue explants or other suppliers of commercially available tissue56,57. One obvious limitation would be that all commercially available tissue models are generated from cell lines in a sterile, "germ-free" environment, meaning the tissue has never been exposed to microbial perturbations: this may exacerbate any inflammatory response in the host, far beyond what would be seen in vivo or following stimulation of ex vivo tissue explants57. Conversely, these laboratory model systems will not contain underlying connective layers or vasculature that one would associate with in vivo tissue, characteristics that can be preserved during ex vivo tissue explantation, and features that impact inflammatory responses. Indeed, a recent systematic review highlighted that careful consideration should be made to utilize tissue models with appropriate vasculature created using various engineering technologies, including biomaterials58. For example, one recent study used a fibrin-based matrix embedded with gingival fibroblasts and microvascular endothelial cells to create a vascularised gingival tissue equivalent. The authors described a differential inflammatory response in the model following exposure to health or disease-associated microorganisms59. From a commercial standpoint, more complex models now exist, such as the "T-Skin model", which contains a full-thickness tissue consisting of a dermis comprised of fibroblasts overlaid with the epidermis. It would be interesting to see how such complex models compare to the epidermis-only systems.
While no model is perfect, organotypic tissue models represent promising alternatives for preclinical testing, aligning with the framework outlined by the three R's, for Replacement, Reduction, and Refinement in undertaking animal research. To this end, although animal models provide important insights into the complex pathophysiological nature of biofilm-related human diseases, they come with obvious disadvantages. These 3D models are easily manipulatable, allowing for large, subtle changes to investigations without ethical approval. Combining these models with existing complex biofilm systems outlined above can greatly improve our understanding of host-pathogen interactions and better predict the success of novel therapies prior to in vivo investigations.