Illuminating the oral microbiome and its host interactions: tools and approaches for molecular ecological studies

Abstract

A more comprehensive understanding of oral diseases like caries and periodontitis is dependent on an intimate understanding of the microbial ecological processes that are responsible for disease development. With this review, we provide a comprehensive overview of relevant molecular ecology techniques that have played critical roles in the current understanding of human oral biofilm development, interspecies interactions, and microbiome biogeography. The primary focus is on relevant technologies and examples available in the oral microbiology literature. However, most, if not all, of the described technologies should be readily adaptable for studies of microbiomes from other mucosal sites in the body. Therefore, this review is intended to serve as a reference guide used by microbiome researchers as they inevitably transition into molecular mechanistic studies of the many significant phenotypes observed clinically.

The review covers molecular ecology approaches to study microbiomes.

Introduction

The major oral diseases, caries and periodontal disease, are the consequences of dysbiotic oral microbial communities, which ultimately develop disease-specific microbial metabolic signatures reflecting the underlying microbial ecology (Nyvad and Takahashi 2020, Zhang et al. 2022). Conversely, the oral microbial community composition can also be impacted by various systemic diseases. For example, the susceptibility to periodontal disease is increased in subjects with diabetes, rheumatoid arthritis, or systemic lupus erythematosus (Graves et al. 2019). An association of oral microbes in extraoral/systemic diseases has been confirmed for inflammatory bowel disease, various types of cancer, and Alzheimer’s disease (Peng et al. 2022). Thus, evidence suggests that the oral microbiome plays a much larger role in systemic health than is commonly assumed, a fact that points to the major importance of the oral microbiome in various extraoral sites in the body. Previous metagenomic analyses of oral microbial communities have already provided detailed inventories of oral microbes, including eukaryotes and bacteriophages, in addition to confirming the association of specific microbial signatures with health and disease (Moussa et al. 2022). Since it is clear that microbial ecology is a critical determining factor for oral health and potentially systemic health as well, ecological investigations have emerged as fundamental components of oral microbiome research. A major goal of ecological studies is to understand the distribution of species in their environmental contexts. A subspecialty in the field, termed molecular ecology, employs the tools of modern molecular biology to reveal how organisms interact with their environments at the molecular level. For such studies, it is crucial to understand which species are present within an ecosystem of interest. It took more than a century to reveal the full extent of the diversity of microbes within the oral microbiome. The early publication “Mycology of the mouth: a text-book of oral bacteria” from K. W. Goadby had already identified >50 different oral bacterial species that were all isolated and characterized using selective culture approaches (Goadby 1903), giving the first glimpses into the species richness in the oral cavity. Culture-independent methodologies have had the greatest influence on our current understanding of human oral microbiome composition. Early culture-independent DNA-based techniques utilized DNA extracted from previously cultivated oral microbes for hybridization in checkerboard assays to determine the microbial composition in healthy subjects and subjects with oral diseases (Socransky et al. 1994). With the development of the first 16S rRNA gene databases, the field quickly adapted this useful resource to improve culture-independent analyses of the oral microbiome. Initially, DNA was extracted from clinical specimens to use as templates for bulk PCR amplification of bacterial 16S rRNA genes, which were then randomly cloned and sequenced using Sanger sequencing. This approach launched the Human Microbiome Project (HMP). The HMP consortium aimed to provide a comprehensive overview of the human microbial communities sampled on 18 human body sites, including the oral cavity (Human Microbiome Project 2012). The oral microbiome now has a dedicated curated database (eHOMD). The subsequent introduction of high-capacity next-generation DNA sequencing technologies revolutionized studies of oral ecology by providing the first comprehensive list of species in the oral microbiome, finally completing a project that began over a century earlier. HMP studies have unambiguously revealed the unique microbial signatures associated with states of oral health and disease. The next great challenge in the field is to understand what these signatures mean and why they exist. For a detailed review about the history of oral microbiome sequencing see (Kumar et al. 2021).

Overview of oral molecular ecology

While a biofilm is composed of many individual microbial cells, it is the higher order community organization that provides the growth benefits to all. Most of the well-characterized bacterial interactions occur within the immediate vicinities of neighboring bacterial cells, where signaling and metabolic communication are most efficient. Many species secrete specific quorum-sensing molecules to communicate with kin, whereas interspecies communications routinely occur via multiple mechanisms, including broadly recognized signal molecules like autoinducer 2 (AI-2) (Swift et al. 1994, Kolenbrander et al. 2002, McNab et al. 2003, Egland et al. 2004), through metabolic exchange, or by directly adhering to other species via evolved physical means like coaggregation/coadherence (Hojo et al. 2009, Kreth and Herzberg 2015). Early studies examining coaggregating pairs of cultivable oral bacteria provided some of the first insights into the broader interaction network and biogeography within oral biofilms (Kolenbrander et al. 2006), while more recent studies have linked the spatial organization of species to states of health vs. disease (Mark Welch et al. 2019, Kim et al. 2020). Even though microbiome 16S rRNA metagenomic studies have already provided the microbial guestlist, further ecological studies are still required to reveal the seating arrangements. Such knowledge is essential to understand how a complex multispecies biofilm develops and functions as a community. This includes studies of the spatial and temporal organization of the oral microbiome as well as its functional outputs like metabolic exchanges. The following section will highlight in vitro, in vivo, and in situ approaches that have all been employed to shed light on these key issues in oral molecular ecology.

Community dining: culture media to recreate and decipher complex microbiome communities

Previous in vitro studies have characterized a variety of interspecies interaction mechanisms between oral bacteria (Egland et al. 2004). Signal molecules, metabolic exchanges, cross-feeding, and physical interactions are all critical determinants of species succession and organization during polymicrobial biofilm development (Darveau and Curtis 2021). Thus, it is clear that reductionist single species assay models are insufficiently complex to reproduce many of these characteristic features of oral ecology. For this reason, it is crucial to have culture media available that can support the development of microcosm communities reflective of those naturally formed clinically within dental plaque or saliva. Using such microcosm cultures, it is then possible to study the influence of specific genes or metabolic pathways on community dynamics. Biologically relevant microcosms are especially useful for modeling the effects of novel therapeutic agents or biomaterials on community composition and/or functionality (Greenman et al. 2020, Eick 2021). In the clinical scenario, saliva serves the principal nutrient source during the development of supragingival oral biofilms, whereas gingival crevicular fluid (GCF) is particularly important for subgingival communities. Fortunately, the composition of both saliva and GCF has been characterized in detail (Carpenter 2013, Barros et al. 2016), which has led to the development of several growth media specifically designed for the culture of oral microcosms. Two such early media, BMM (modified basal medium containing mucin) and DMM (defined medium containing mucin), were designed based on the composition of saliva and were demonstrated to sustain growth of a diverse microbiome in a controlled environment (Wong and Sissons 2001). More recently, another medium called SHI medium (named after the senior author of the study) was developed to support an even greater diversity within oral microcosms (Tian et al. 2010, Edlund et al. 2013). SHI medium is noteworthy because it incorporates a variety of important host-derived components known to influence the growth of different groups of oral bacteria, a strategy that can be easily adapted for other host-associated niches. 16S rRNA analyses of SHI medium-derived microcosms found that they contained 60% – 80% of the original operational taxonomic unit (OTU) richness found in the original inoculum. The inoculum source (pooled saliva from different subjects) could account for some of the variation encountered in OTU richness (Edlund et al. 2013). When developing a new microcosm growth medium, it is important to note that the input clearly determines the output. If the composition of the initial community is already known through metagenomic sequencing or by other means, key pathways can be modeled to identify important auxotrophies that could be complemented in the medium composition (Zengler and Zaramela 2018, Johnson et al. 2020, Seif et al. 2020). For example, certain rare carbohydrates may be added to a recipe if essential biosynthetic or catabolic pathways are missing from one or more members of the community. Since the growth of one or more bacterial species can affect numerous others within an interdependent microbial ecosystem, complementing the growth of one organism can have an outsized effect on the overall diversity obtained in a microcosm community. This point was nicely illustrated in a recent study of previously uncultivated subgingival bacteria. Subgingival dental plaque specimens were loaded into wells formed in agar plates supplemented with siderophores to facilitate iron uptake, an essential nutrient for most bacteria. Bacteria commonly sequester free iron using secreted siderophores, which become common goods in a community setting, facilitating iron uptake from other species lacking high affinity uptake systems (Saha et al. 2013). This simple alteration to the culture conditions successfully supported growth of previously uncultivable subgingival phylotypes, including members from Spirochaetes, Actinobacteria, Bacteroidetes, Firmicutes, Fusobacteria, Proteobacteria, and Synergistetes. It also supported the cultivation of a novel taxon, Anaerolineae bacterium HOT-439, the first member of the phylum Chloroflexi from the human oral cavity (Vartoukian et al. 2016). The aforementioned SHI medium similarly supported the cultivation of previously uncultured interdependent communities, including members of the phylum Saccharibacteria (initially designated TM7). One such organism, Nanosynbacter lyticus strain TM7x, is a severely genome-reduced obligate epibiotic parasite of Actinomyces odontolyticus and is the first cultivated member of the phylum Saccharibacteria. This organism was originally cultivated in SHI medium together with its host A. odontolyticus. Since the first report of its cultivation in 2015 (He et al. 2015), several follow-up studies of N. lyticus have contributed significantly to our understanding of its host range, host interactions, and association with periodontal disease (Bedree et al. 2018, Bor et al. 2019, 2020, Schulz et al. 2021). Lastly, it is worth noting that any newly developed culture medium yielding the desired community diversity may still perform poorly for sustaining community viability (Baraniya et al. 2020). Therefore, additional additives may be required to further improve the overall performance and utility of a microcosm growth medium.

Research into the microbiome secretome is another potential application of microcosm community culture. Little is currently known about the oral secretome, but these molecules are likely to exert profound impacts on community development and may even impact microbiome–host interactions. For example, many bacteria secrete toxic molecules designed to attack other competitor species in the microbiome (Kreth et al. 2011), while others secrete molecules to specifically modulate host responses (Okamura et al. 2021). SHI medium-derived microcosm cultures were used to create complex communities subsequently analyzed via liquid chromatography-tandem mass spectrometry (LC-MS/MS) to identify secreted peptidic small molecules in the cultures (Edlund et al. 2017). Not only did this study identify hundreds of new molecules, it also revealed that the secretome is dynamic, evolving temporally much like the community itself. Furthermore, certain enzymatic modifications such as acetylation and methylation were shown to contribute to the diversity of the detected peptidic small molecules (Edlund et al. 2017). Certainly, we are only just beginning to understand how the microbiome secretome influences oral microbial ecology and host interactions. It is conceivable that the secretome would also influence host systemic health, since these molecules are ultimately transported into the gastrointestinal tract through normal processes. At sites of active oral disease, secreted molecules can be directly injected into the bloodstream, which may further influence the systemic health of the host (Zhang et al. 2020).

Biofilm extracellular polymeric substances (EPS)—deciphering its ecological role through fluorescent staining and biophysical characterization

EPS fluorescent imaging approaches

Human-associated biofilm communities not only consist of microbial cells, but also contain a diverse assembly of extracellular components. This matrix of extracellular polymeric substances (EPS) is a defining feature of biofilms. EPS is typically a complex mixture of carbohydrates, proteins, nucleic acids, lipids, cell wall-derived peptidoglycan (Karygianni et al. 2020), and may even include a variety of host molecules (Alhede et al. 2020). The exact composition and specific type of EPS are both determined by the biofilm constituents and may vary depending on time and location. While EPS provides the structural scaffolding of a biofilm, it also plays critical roles in antibiotic protection, cell–cell signaling, nutrient flow/waste removal, genetic exchange, and it sometimes supports additional niche-specific functions (Costa et al. 2018). When considering microbiome–host interactions, the EPS is one of the principal microbial components that directly interacts with epithelial and immune cells, and it plays a central role in oral diseases (Bowen et al. 2018). Although EPS is an integral component of the biofilm, there is a surprisingly limited understanding about its specific composition or dynamics, especially within the host setting. Much of what is currently known has been derived from single species in vitro biofilms. This approach simplifies the isolation of sufficient EPS quantities for downstream compositional analyses. However, EPS dynamics are substantially more complex within the oral cavity and additional research is required to better understand its ecological role in the human microbiome. Dental plaque EPS has been recently reviewed (Jakubovics et al. 2021).

In situ imaging studies of EPS material can be quite challenging due to both the complexity of EPS composition as well as the myriad of molecular interactions occurring within. EPS can be extracted and purified for downstream analyses using chemical and physical techniques like centrifugation, ultrasonication, EDTA treatment, and alkaline extraction (Di Martino 2018). EPS chemical composition can be deciphered with techniques such as Fourier transform infrared (FT-IR) spectroscopy or other chemical approaches (Di Martino 2018). EPS structure is often characterized using fluorescent probes, as was previously demonstrated for the extracellular DNA (eDNA) found within bacterial biofilms, including those of the oral cavity (Serrage et al. 2021). eDNA seems to play a critical function in most biofilm development by providing substantial structural support in addition to a variety of additional ecological functions. (Roberts and Kreth 2014). Both Gram-positive and Gram-negative oral bacteria have been shown to excrete eDNA through active mechanisms (Serrage et al. 2021). It is currently unclear whether biofilms in the oral cavity also contain host DNA, but there is reason to suspect that it would, especially for biofilms formed in the presence of chronic inflammation. For example, abundant host DNA is found within the biofilms of chronic lung infections, where substantial amounts of eDNA are released through NETosis and by necrotic neutrophils (Alhede et al. 2020). The lattice-like structure of the eDNA within the EPS of these pulmonary biofilms was revealed by fluorescently labeling with an α-dsDNA primary antibody in combination with an Alexa Fluor 488-labeled secondary antibody. Presumably, this strategy would be adaptable to other microbiome-associated biofilms as well. Recent evidence indicates that the eDNA structure observed in biofilm EPS is directly mediated by extracellular DNABII family proteins (integration host factor, IHF, and histone-like protein, HU) (Devaraj et al. 2019). These proteins exhibit high affinities for branched DNA, such as those found in Holliday junctions, and they can remodel the eDNA once they bind. Importantly, DNABII family proteins are responsible for converting the typical B-form eDNA excreted in biofilms into a DNase-resistant Z-form (Buzzo et al. 2021). Z-form eDNA was shown to be a common constituent of a broad cross-section of bacterial biofilms, including those of Streptococcus mutans, and it can be created from eDNA released by both bacterial and host cells. Thus far, DNABII family proteins have been identified within the EPS of a variety of oral bacteria, including S. gordonii, S. mitis, S. mutans, S. oralis, and Porphyromonas gingivalis (Rocco et al. 2017, 2018, Devaraj et al. 2019, Buzzo et al. 2021).

Biofilms may also contain extracellular RNA (eRNA), but it is not yet clear what role it plays in the biofilm matrix. eRNA was identified in the biofilms of Candida albicans (Smolarz et al. 2021), a fungal species commonly found in biofilms of the oral cavity and upper respiratory tract. Specific dyes are available that can discriminate between eDNA and eRNA (Table 1). Fluorescent probes are also available to label EPS carbohydrates, which often comprise substantial fractions of the total biofilm EPS. In the oral cavity, supragingival dental plaque is an especially rich source (Jakubovics et al. 2021). Some of these molecules have been targeted for fluorescent labeling. For example, the extracellular glucan polymers synthesized by S. mutans can be specifically labeled using fluorescent dextrans, which become incorporated into the glucan polymer network when added to developing S. mutans biofilms (Klein et al. 2009). Carbohydrates are also commonly present in glycoconjugates covalently linked to proteins, lipids, and various small molecules. These have been successfully labeled in oral biofilms using fluorescence lectin-binding analysis (FLBA) (Neu et al. 2001, Neu and Lawrence 2014, Tawakoli et al. 2017). Lectins are proteins that preferentially recognize and bind to specific carbohydrate complexes found on glycoconjugates, and they can be used in an analogous manner to antibodies (Ghazarian et al. 2011). Therefore, FLBA can reveal both the structural distribution of carbohydrates within biofilms as well as their composition. A comprehensive study examining 75 fluorescently labeled lectins were screened for their abilities to label matrix components of pooled oral biofilms and a variety were found to confer widespread fluorescence (Table 1). In addition, FLBA labeling can be employed to analyze the biovolume of biofilm glycoconjugates, particularly when paired with fluorescence in situ hybridization (FISH) of the bacterial cells (see below). This combination provides a detailed overview of bacterial distribution together with the carbohydrates they produce as part of the surrounding biofilm EPS (Seviour et al. 2019). It is worth noting that fluorescent lectin labeling can sometimes fail to discriminate between EPS matrix material and other biofilm components. For example, certain lectins can preferentially label cell surface carbohydrates, yielding fluorescent outlines of individual cells, rather than creating the cloud-like appearance, i.e. typical of lectin-labeled EPS (Neu et al. 2001). For this reason, it is advisable to include cell counterstains and/or dispersed biofilms as controls. Human clinical biofilms may also show considerable variation with lectin staining (Tawakoli et al. 2017). This may be due to problems of diffusion through thick biofilms or from competitive inhibition by free monosaccharides in the specimen. These issues can be particularly problematic for quantitative analyses of lectin-stained images.

Table 1.

Examples of specific fluorescent staining methods for EPS.

EPS	Detection method	Reference
eDNA	YOYO-1 TOTO-1 Sytox Green α-dsDNA monoclonal antibody, requires fluorescent secondary AB Quant-iT PicoGreen; specific for double stranded DNA	Rostami et al. (2017) Schlafer et al. (2017) Nagasawa et al. (2020) Devaraj et al. (2019) Smolarz et al. (2021)
eRNA	SYTO RNASelect	Smolarz et al. (2021)
Carbohydrates	Calcofluor white M2R for β-polysaccharides Lectins (examples): AAL (Aleuria aurantia) ⇒ α-fucoseCalsepa (Calystega sepiem) ⇒ mannose and maltoseHPA (Helix pomatia) ⇒ N-acetyl-α-galactosamineLEL (Lycopersicon esculentum) ⇒ N acetylglucosamineMNA-G (Morniga-G) ⇒ galactoseMany others available, see Lectin Frontier Database Fluorescently labeled glucans primers	Gonzalez-Machado et al. (2018) Hirabayashi et al. (2015), Tawakoli et al. (2017)
Proteins	FilmTracer™ SYPRO® Ruby Biofilm Matrix Stain Epicocconone	Ali Mohammed et al. (2013), Randrianjatovo et al. (2015)
	Amyloid fibers: WO1 for amyloid proteins, requires fluorescent labeling of the specific antibody CDy11 Congo Red Thioflavin T	Larsen et al. (2007) Kim et al. (2016) Erskine et al. (2018) Erskine et al. (2018)
Lipids	NileRed DiD'oil	Rumin et al. (2015) Gonzalez-Machado et al. (2018)
Others	Vancomycin-BODIPYTM (vancomycin binds to peptidoglycan from Gram-positive cells, thus stains intact cells, but also free peptidoglycan)	Daddi Oubekka et al. (2012)

Extracellular matrix material is also composed of a variety of proteins and lipids (Karygianni et al. 2020), and these too can be fluorescently labeled within biofilm EPS (Table 1). Fibrillar proteins (also referred to as amyloid-like fibers) are commonly found in the matrix of bacterial biofilms (Erskine et al. 2018). They play a significant role in S. mutans biofilm formation (Oli et al. 2012), where some appear to form complexes with eDNA (Chen et al. 2019). Amyloids are commonly visualized using the fluorochrome congo red, which can be directly added to biofilms for imaging studies (Chen et al. 2019).

In addition to its structural role in biofilms, EPS shapes a variety of biofilm microprocesses like cell–cell communication, metabolic interactions, and structural remodeling. These microprocesses are crucial for the proper maintenance of the biofilm community and can even orchestrate community-wide responses to environmental stress (Costa et al. 2018). Several imaging techniques have been applied to monitor such events, including mass spectrometry imaging, confocal laser scanning microscopy (CLSM), Raman spectroscopy imaging, and atomic force microscopy (AFM; Ivleva et al. 2017, Zhang et al. 2019). Fluorescent pH nano-sensors have been successfully employed to image pH gradients formed by diffusion through biofilm EPS of both S. mutans and P. aeruginosa (Hollmann et al. 2021).

Biophysical investigations of the microbiome

The biophysical properties of biofilm EPS have yet to receive much attention in the field, presumably due to the lack of biofilm-specific test methods available. Several technologies have been adapted for the study of biofilm mechanical properties, including oscillating rheology, microindentation, and microjet impingement (Gloag et al. 2020). A total of 30 years ago, cell-free glucans isolated from S. mutans were first investigated for their rheologic behavior in the presence of saliva (Rundegren et al. 1991). However, biofilm mechanical properties remained largely uncharacterized until work from the Stoodley laboratory demonstrated that intact S. mutans biofilms could be analyzed using a rheometer, an instrument, i.e. typically employed in the material science field for the study of mechanical stress resistance. It was discovered that under shear stress conditions, S. mutans biofilms exhibit viscoelastic properties similar to other rheologic fluids (Vinogradov et al. 2004). The initial application of rheologic measurements laid the groundwork for later studies aimed at understanding the genetic factors controlling S. mutans biofilm mechanical properties (Palmer et al. 2019, Marx et al. 2020). In one such study, rheological measurements were crucial for demonstrating how S. mutans actively remodels its biofilm viscoelastic properties when challenged with environmental stress (Marx et al. 2020). This process is regulated by the stress-sensitive transcriptional repressor IrvR and a downstream component of its regulon, the cell surface lectin GbpC. Under nonstress growth conditions, GbpC is produced in low abundance, which is required for normal biofilm formation due to its affinity for the glucan polymers in the biofilm EPS (Lynch et al. 2007, Liu et al. 2015). However, in the presence of a variety of environmental stresses, cell surface GbpC production among biofilm cells is dramatically increased, resulting in a network of additional GbpC-glucan interactions in the EPS that become “knitted” to create a rigid biofilm structure (Marx et al. 2020). While this mechanism was demonstrated using single species S. mutans biofilms, it is conceivable that it could similarly affect mixed species communities as well because GbpC is able to cross-link the exopolysaccharide matrix to other oral biofilm members, especially in the presence of environmental stress (Koo et al. 2013). It will be interesting to determine in future studies whether environmental conditions can trigger a remodeling of the viscoelastic properties of polymicrobial biofilms and if certain species may be involved in the process.

Microindentation and microjet impingement have been similarly employed to investigate the mechanical properties of biofilms at the microscale. Microindentation uses a mechanical indenter controlled by a force transducer, allowing precise penetration of the indenter into the biofilm with subsequent lateral movements to evaluate mechanical properties and failure mechanisms (Cense et al. 2006). Like the lateral lesions measured by microindentation, microjet impingement also employs shear force to create biofilm lesions. In this case, a precisely controlled stream of water is responsible for creating the lesion. By measuring the applied water pressure, delivery time, and lesion size one can calculate the lateral shear force required to detach a biofilm from the surface (Deshpande 1983). This approach was used to correlate S. mutans glucan concentration in the EPS with the shear resistance of its corresponding biofilms, demonstrating an exponential relationship (Kreth et al. 2004a).

AFM

AFM, also referred to as scanning probe microscopy (SPM), is a probe-based nonoptical imaging technique that can be used to quantify several biologically relevant surface parameters. While initially developed to characterize semiconductor surfaces, it was quickly adapted for high-resolution biophysical studies of oral biofilm EPS and surface interactions (Sharma et al. 2010, Bhat et al. 2021). AFM employs a cantilever-mounted tip, i.e. moved over a biological surface, such as a biofilm, with the help of a piezoelectric micropositioner. Physical interactions between a surface and the cantilever tip trigger distortions in the tip that can be precisely measured by the deflection of a laser beam focused above the cantilever (Dufrene 2008). AFM can be operated in several different modes, including contact mode, noncontact-mode, and tapping mode. Tapping mode is especially useful to measure mechanical properties, such as the elasticity and hardness of microbial cells or entire biofilms (James et al. 2017). As a proof of principle, we originally employed AFM to investigate the structure–function relationships of different S. mutans isolates and mutant strains (Cross et al. 2006). In a follow-up AFM study, the S. mutans cell wall protein WapA was specifically investigated to demonstrate its role in cell–cell aggregation and biofilm formation (Zhu et al. 2006). Several studies have employed functionalized silicon or silicon nitride AFM cantilever tips to measure cell–cell interaction forces. For example, cross-kingdom interactions between S. mutans and C. albicans have been studied using this approach (Wang et al. 2020, Wada and Nomura 2021, Wan et al. 2021). Extracellular glucan production by S. mutans was shown to selectively favor S. mutans interactions with C. albicans during cariogenic biofilm development, which has important implications for the frequent clinical association of these two organisms in children experiencing Severe Early Childhood Caries (SECC) (Wan et al. 2021). Even though only a relatively small number of studies have investigated biofilm biophysical properties, it is already clear that such properties are more dynamic than previously assumed. Considering the importance of biofilm formation for most human microbiome-associated diseases (Curtis et al. 2020, Welp and Bomberger 2020), this area is likely to receive significant attention in the coming years.

Hi neighbor, who are you? Fluorescent imaging techniques

As previously described, the ecological aspects of oral disease were first recognized many decades ago. Consequently, there has been a longstanding interest in the field to understand the spatial relationships between different groups of organisms within the oral biofilm. Early studies in this area employed in vitro coaggregation assays between pairs of species to infer the broader arrangement of organisms in polymicrobial biofilms, ultimately created via adhesin interactions between different organisms (Rickard et al. 2003, Kolenbrander et al. 2006). Subsequent studies revealed that oral biofilms develop in distinct and reproducible waves of bacterial colonization, leading to the broad classification of oral bacterial species into pioneer/early colonizers, bridge species, and late colonizers. Accordingly, polymicrobial biofilm development was shown to develop in a highly biased orderly fashion, rather than through a random accretion of oral bacteria. Even though recent studies have challenged some of the previous assumptions inferred from coaggregation studies (Kolenbrander et al. 2006), remarkably, we now know that this traditional view of oral ecology is fundamentally accurate, due in large part to studies employing fluorescence in situ hybridization (FISH) (Mark Welch et al. 2019, Borisy and Valm 2021).

Fluorescence in situ hybridization

FISH was originally developed to localize specific genetic sequences in Xenopus and HeLa cells using hybridized radioactive probes complementary to target DNA or RNA in the cells (John et al. 1969, Pardue and Gall 1969). With the subsequent development of distinct fluorescently labeled probes, FISH was adapted and further developed as a microbiological technique to detect, identify, and enumerate microorganisms directly within environmental and clinical samples using fluorescence microscopy applications like CLSM. The pairing of FISH labeling with CLSM has played a particularly important role in our current understanding of the spatial organization of oral biofilms because it places the taxonomic distribution of microbes into a 3D context. Identified species interaction networks can then be specifically investigated at the biochemical and genetic levels. The taxonomic specificity of FISH is dependent on the design of fluorescent probes complementary to the ribosomal RNA of target microbes (DeLong et al. 1989). These probes can be designed with specificities capable of discriminating between individual species up through higher taxonomic ranks. As the vast majority of RNA in microbial cells is composed of rRNA, this creates a reliable pool of abundant targets for hybridization and guarantees high fluorescent signal intensities suitable for imaging. The de novo design of effective 16S rRNA FISH probes (or 18S rRNA for eukaryotes) requires the careful consideration of multiple factors, including probe length, cross-reactivity, melt temperature, G + C content, secondary structure, and target accessibility within the rRNA (Yilmaz and Noguera 2004, Lima et al. 2020). Several web-based resources are available to aide in the design of new FISH probes, with experimental FISH optimization protocols available in several excellent publications (Moter and Gobel 2000, Yilmaz et al. 2011, Karygianni et al. 2014, Teixeira et al. 2021). In many instances, the desired FISH probes have already been created and verified in previous studies, and these can be identified using web-based databases like probeBase (https://probebase.csb.univie.ac.at/).

Fluorescence in situ hybridization of oral biofilms

One of the early studies employing FISH with oral biofilms investigated a six-species in vitro biofilm model (Thurnheer et al. 2004). While this was not the first FISH study to examine multispecies oral biofilms (Wecke et al. 2000), its development of protocols for simultaneous FISH imaging of Gram-positive and Gram-negative bacteria together with fluorescent EPS staining provided a template that influenced later studies in the field. FISH has proven to be especially important for the study of clinical oral biofilms sampled directly from human volunteers. The oral cavity is highly amenable to clinical research, which provides intriguing opportunities to examine biofilm ecology using removable intraoral appliances. These devices are worn by volunteers and house multiple removable tooth enamel slabs that serve as clinically relevant substrates for biofilm formation. In one such study, volunteers wore the intraoral appliances for a 7-day time course, with individual enamel slabs being removed on days 1, 2, 3, 5, and 7 followed by subsequent FISH analyses of the intact, unaltered biofilms. Consistent with previous models of oral biofilm development, CLSM FISH images revealed a clear succession of species over the time course as the biofilm increased in thickness. Streptococcus species remained highly abundant throughout, but decreased significantly by day 7. The pioneer colonizer Actinomyces naeslundii steadily decreased in abundance as the biofilm matured, whereas the organic acid-fermenting Veillonella species remained relatively constant. The ubiquitous bridge species Fusobacterium nucleatum decreased in representation at early time points, but became quite prevalent with biofilm growth and maturation. The number of species examined in this study was inherently limited by the number of unique fluorescent signals that could be simultaneously discriminated. By comparing the signal abundance ratios from these taxon-specific probes compared to the broad domain-specific probe Eub388 (Amann et al. 1990), the authors estimated that 40%–65% of the total biomass was represented by other undetected species. In agreement with previous coaggregation research, this study provided direct evidence for the nonrandom distribution of oral microbes in vivo (Al-Ahmad et al. 2007). With the subsequent development of additional unique fluorophores, a larger number species could be simultaneously labeled in oral biofilms. However, due to the relatively narrow range of wavelengths available for fluorescence microscopy, technical limitations are imposed by the overlapping excitation and emission spectra of many fluorophores. This greatly restricts the number of distinct taxa that can be simultaneously discriminated. To address this limitation, a technique was developed called combinatorial labeling and spectral imaging (CLASI) with FISH or CLASI-FISH (Valm et al. 2011). The CLASI-FISH approach labels individual microorganisms with two or more distinct fluorophores that can be unmixed computationally to discern organisms having each predetermined combination of fluorescent labels. An early application of CLASI-FISH gave a hint of its future potential by demonstrating how only 6 fluorophores could support imaging 15 unique organisms (Valm et al. 2012). Soon after, this theoretical upper limit was expanded to discriminate 120 distinct microbe labels using just 16 commercially available fluorophores (Valm et al. 2016). Considering that the typical individual harbors up to 300 distinct bacterial taxa in the oral cavity (Darveau and Curtis 2021), CLASI-FISH is the now at the point where a systems-level analysis of species interactions is a realistic endeavor. An early study in this area investigated the spatial relationship and species abundance of semidispersed human dental plaque specimens using 15 human oral taxon-specific probes. Spatial arrangement information obtained from CLASI-FISH resulted in the identification of 36 significant interspecies associations within the cell aggregates (Valm et al. 2011). Using techniques like CLASI-FISH, it is currently possible to develop micron-scale biogeographical landscapes with quantitative analyses of species distributions (Mark Welch et al. 2016, Wilbert et al. 2020, Borisy and Valm 2021). This is providing a far more nuanced understanding of the polymicrobial community as a whole. For example, in a seminal study combining metagenomic sequencing and FISH analysis of supragingival oral biofilms, Mark-Welch et al. (2016) identified the Corynebacterium genus as a highly abundant foundational oral biofilm taxon. FISH images of filamentous oral corynebacteria revealed interspecies adherence by oral streptococci, which directly bind to these filaments in so-called corncob structures (Mark Welch et al. 2016). Based on these results, we further investigated the phenotypic consequences of dual species cultures of the oral Corynebacterium species C. durum and the commensal oral Streptococcus species S. sanguinis. C. durum was found to excrete fatty acid-containing membrane vesicles that induced both morphological and physiological changes in S. sanguinis, increasing its fitness (Treerat et al. 2020). FISH analysis similarly revealed the ecological basis for the frequent association of two highly abundant oral commensal species, S. mitis and Haemophilus parainfluenzae (Perera et al. 2021) (Fig. 1A and B). Like many oral streptococci, S. mitis is a robust hydrogen peroxide producer, creating enough to inhibit the growth of H. parainfluenzae (Redanz et al. 2018). However, S. mitis also excretes NAD, which complements the NAD auxotrophy of H. parainfluenzae. Consequently, H. parainfluenzae has developed a multifactorial oxidative stress response to cope with S. mitis hydrogen peroxide production. Even so, H. parainfluenzae still avoids direct contact with high density clusters of S. mitis, presumably to avoid overloading its oxidative stress resistance capacity (Perera et al. 2021). Besides CLASI-FISH, a couple additional modifications to the FISH technique are worth mentioning. The first is peptide nucleic acid FISH (PNA FISH). PNA FISH probes utilize fluorescently labeled peptide nucleic acid DNA mimics. These probes can be made shorter than traditional oligonucleotide-based FISH probes, and yet, they exhibit enhanced uptake and binding kinetics for improved performance (Cerqueira et al. 2008). FISH has also been combined with flow cytometry (FLOW-FISH) as a quantitative approach to analyze specific organisms within biofilm communities (Pereira et al. 2021). Catalyzed Reporter Deposition or CARD-FISH combines the specificity of nucleic acid hybridization with the signal amplification provided by horse radish peroxidase (HRP) instead of a fluorophore, thus enabling the detection of low abundance species. A specific substrate for HRP, tyramide conjugated with fluorescent dyes, can then be rapidly converted to strong fluorescent signals (Kubota 2013). Other FISH techniques are also available, and like CARD-FISH have not been used for the investigation for oral biofilm communities, but hold a great potential for deeper exploration, including MAR-FISH and NanoSIMS-FISH. For a more comprehensive overview see (Guimarães et al. 2021).

Figure 1.

Example of fluorescent in situ hybridization of dental plaque samples to support in vitro phenotypic observation. The FISH-probes used were specific for Haemophilus spp. (probe Hpar441; Haemophilus spp.; cyan) and S. mitis group species (probe Smit651; mitis group streptococci; magenta). The picture in (A) shows plaque with sparsely distributed Haemophilus spp. among a majority of S. mitis group bacteria. Picture (B) represents high abundance of both species. The magnification in (i) shows that most Haemophilus spp. cells are within a few microns of the nearest S. mitis group cell. Generally, Haemophilus spp. cells seem to be well intermixed. (ii) Haemophilus spp. avoids the highest densities of the S. mitis group due to the elevated H₂O₂ production, i.e. toxic to Haemophilus spp.; see text for details. Scale bars indicate 10 µm. Picture was adapted from reference (Perera et al. 2021), with permission.

With the rapid evolution of FISH studies, the field has already advanced beyond the initial question of: Hi neighbor, who are you? and we are now beginning to address the more compelling question: Hi neighbors, why are you here?

Imaging with fluorescent proteins

Labeling cells with FISH probes or nonspecific fluorescent dyes, like membrane, DNA, or protein dyes, provides a mechanism to decipher biogeography, biofilm architecture, and the temporal development of microbiome communities. However, these labels are less useful for deciphering dynamic cellular processes such as alterations in gene expression within the microbiome community. Fluorescent reporter proteins like the green fluorescent protein (GFP) have long been employed for such studies in single species settings (Southward and Surette 2002), and they are becoming more commonly employed in multispecies biofilm studies as well (Gu et al. 2005, Shields et al. 2019). Fluorescent reporter genes can provide valuable insights that connect larger ecological trends to individual bacterial gene expression. For example, fluorescent proteins have long been employed to investigate a number of oral streptococcal ecological functions like bacteriocin production and natural competence using both single and multispecies biofilms (Kreth et al. 2004b, 2005, Merritt and Qi 2012, Qin et al. 2021). Furthermore, fluorescent reporters have played central roles in revealing how the spatial arrangement of oral bacteria impacts genetic responses in the oral biofilm. A GFP transcription fusion to the S. gordonii amylase-encoding gene amyB demonstrated that its expression is specifically activated by interspecies cell–cell communication with Veillonella atypica, but only when the two organisms are in close proximity. This interaction increases carbohydrate utilization by S. gordonii and results in a corresponding increase its excretion of the toxic metabolic waste product lactic acid. Lactic acid is also the preferred carbohydrate source of V. atypica (Egland et al. 2004). Thus, both organisms grow more efficiently as a pair than as individuals. Fluorescent reporter strains of S. gordonii have been similarly employed to reveal spatially resolved interactions with another microbiome member Aggregatibacter actinomycetemcomitans. Like V. atypica, A. actinomycetemcomitans metabolizes lactic acid excreted by S. gordonii. However, it is also sensitive to the hydrogen peroxide naturally produced by S. gordonii during carbohydrate catabolism. Thus, when hydrogen peroxide production outpaces the detoxification capacity of A. actinomycetemcomitans, this will trigger the production of glycoside hydrolase dispersin B to escape the biofilm and ultimately maintain an appropriate distance that balances the benefits of lactic acid acquisition with the challenges of oxidative stress (Stacy et al. 2014). The disadvantages for fluorescent protein reporter studies are that they require genetic manipulation of organisms as well as available molecular oxygen to support fluorochrome assembly. The requirement for genetic manipulation precludes fluorescent protein reporter studies of in situ biofilms. This is a major distinction with FISH studies, which support fluorescent labeling without requiring genetic manipulation or oxygen. For ecology studies, it is also important to confirm that reporter gene expression does not interfere with strain fitness. It is worth noting that the previously described reporter proteins LacZ (beta-galactosidase) and GusA (beta-glucuronidase) both have chemiluminescent substrates (Olesen et al. 2000). Thus, they could conceivably be employed as alternatives to proteins like GFP.

Biophotonic imaging in animal models

The development of the oral microbiome is influenced by a myriad of factors from the environment, host behaviors, host genetics, and the immune response (Kreth et al. 2020). To understand many of the molecular mechanisms involved in oral health, it is necessary to include a complex growth environment like that of the human oral cavity. As discussed above, the requirement for genetic manipulation renders reporter strains inappropriate for use in human clinical research. Thus, one is limited to either in vitro model systems or animal models. Current in vitro model systems lack the complexity required to replicate many of the unique aspects of the host oral environment. Therefore, animal model systems still offer the best options to address certain mechanistic questions about oral ecology. Mice, rats, and primates have all been employed for this purpose (Genco et al. 1998). While these studies have revealed numerous insights into the microbial and host responses to the infections, spatial information about the distribution of test organisms or their in situ gene expression patterns are largely lacking. Animal experiments are typically endpoint-based and require invasive procedures to enumerate the infecting organisms. This precludes temporal studies within individual animals, limiting the scope of ecological questions that can be addressed in vivo. A suite of optimized luciferases has been specifically adapted to address this issue using in situ biophotonic imaging approaches (Merritt et al. 2016). These luciferases were selected based on their unique substrate requirements and emission spectra, which provide a variety of options for multiplex imaging studies within single animals. Using luciola red luciferase-labeled S. mutans and green renilla luciferase-labeled S. gordonii, dual-species oral infections of mice were noninvasively quantified, with separate luciferases each providing both spatial and temporal resolution over a 14-day period (Merritt et al. 2016). This approach requires the use of whole animal imagers like IVIS® and In-Vivo Xtreme^TM (Fig. 2A). The In-Vivo Xtreme instrument supports both optical and X-ray imaging, making it possible to directly observe the precise locations of reporter signals emanating from the oral cavity. In the example shown in Fig. 2(B), S. mutans luciferase activity can be unambiguously localized to a specific mouse mandibular molar (Fig. 2B). By expressing two separate luciferases in a single strain, it was possible to adapt this system to investigate the role of S. mutans bacteriocin gene expression during growth in a complex community in vivo. The S. mutans bacteriocin mutacin IV was placed under the control of a xylose-inducible expression system followed by a green renilla luciferase reporter to measure mutacin IV gene expression. Additionally, this strain concurrently produced a constitutive luciola red luciferase reporter to serve as an internal control for normalization purposes. By modulating the presence of xylose inducer in the mouse drinking water, it was possible to noninvasively measure the corresponding changes in mutacin IV gene expression as well as its impact on S. mutans persistence within the mice oral cavities (Merritt et al. 2016). Unlike S. mutans, other oral streptococci like S. gordonii utilize hydrogen peroxide production as their principal strategy for maintaining a competitive growth advantage (Redanz et al. 2018). This point was similarly demonstrated using biophotonic imaging in mice. By altering the levels of magnesium provided in the mice drinking water, it was possible to exogenously control hydrogen peroxide production from bioluminescent S. gordonii introduced into the mice. Over a 14-day assay period, the abundance of S. gordonii in the mice correlated with the levels of hydrogen peroxide produced, further illustrating the importance of offensive weaponry during growth competition in vivo (Cheng et al. 2020). Biophotonic imaging approaches can provide many new options to explore mechanistic questions of microbiome ecology, and these are only just beginning to be explored. Furthermore, it is feasible to overlay biophotonic imaging data with other potentially useful technologies like micro-CT or positron emission tomography (PET). Indeed, some whole animal imagers have these capabilities already integrated, which would make it feasible to correlate bacterial growth with particular oral disease outcomes like bone resorption.

Figure 2.

Examples of biophotonic imaging using bioluminescent-tagged oral streptococci modeling a realistic oral environment in rodents. (A) Dual-species infection of mice with S. mutans and S. gordonii firefly and renilla luciferase reporters. Composite overlay images of the X-ray and bioluminescent images of the dual infections 6- and 14-days postinoculation demonstrating a stable infection of the mouse with both species over a longer time period. (B) Spatial resolution of mice orally infected with green renilla luciferase reporter-tagged streptococci. Mice were imaged from the top and side to localize the quadrants of infection. The arrows in the X-ray image depict the positions of the maxillary and mandibular molars. The presented images are of a single representative mouse in which the sites of colonization can be localized to both of the mandibular molars. Picture was adapted from reference (Merritt et al. 2016), with permission.

Scanning electrochemical microscopy

Scanning electrochemical microscopy (SECM) is an electrochemical probe-based analytical technique, i.e. able to noninvasively map the chemistry of the 3D space above a biofilm. Detailed descriptions of the electrochemistry principles supporting SECM can be found elsewhere (Amemiya et al. 2008, Huang et al. 2018). Briefly, a low current potentiostat-controlled ultramicroelectrode probe, i.e. selective for a specific charged chemical signature (e.g. pH, Ca²⁺, and so on) is scanned across a substrate such as a biofilm with the help of a high-resolution 3D positioning system. Computer assisted analysis can then translate these data into a 3D map of the measured chemical, resulting in a topographic image. SECM can be performed in several different modes, including feedback mode, potentiometric mode, and redox competition mode to characterize different biological processes. Microelectrodes vary in thickness between 5 and 25 µM in diameter, but they can also be custom-made at even smaller diameters using quartz capillaries (Elsamadisi et al. 2011). SECM has the major advantage of yielding real-time measurements of bacteria or biofilms even when chemical microenvironments are changing in response to evolving environmental variables, providing a kinetic assessment of the bacterial responses to different stimuli. SECM can also effectively measure chemical gradients established over very short distances, whereas other techniques tend to measure bulk solutions, thus averaging these gradients to yield a single value within a heterogenous sample. Currently, SECM is limited by the relatively small number of chemicals that can be measured, as individual SECM probes must be created for each unique chemical signature of interest. Single and dual channel electrodes have been used to characterize several single species biofilms (Abucayon et al. 2014, Harris et al. 2016). A recently developed SECM hydrogen peroxide probe was employed to measure the hydrogen peroxide concentration gradients produced by S. gordonii biofilms (Liu et al. 2011). These gradients could be modified by the introduction of the catalase-producing species A. actinomycetemcomitans, illustrating the importance of community composition for S. gordonii hydrogen peroxide production (Liu et al. 2011). Similarly, in a dual species artificial saliva community of S. gordonii and S. mutans, pH and hydrogen peroxide SECM probes demonstrated S. mutans ability to inhibit hydrogen peroxide production by neighboring S. gordonii cells, despite it lacking any known hydrogen peroxide-degrading enzymes. It was later determined that increasing acidification of the local environment by S. mutans represses S. gordonii expression of the pyruvate oxidase-encoding gene spxB, the main enzyme responsible for its competitive production of hydrogen peroxide (Joshi et al. 2017, Cheng et al. 2018). Thus, aciduricity represents only one part of the competitive growth advantage afforded to S. mutans in acidic environments containing S. gordonii. SECM has also been combined with other technologies like micro-3D printing, which was used to spatially arrange bacterial populations into 3D printed micro “lobster-traps” to study cell–cell communication (Connell et al. 2014). When combined with AFM, SECM is able to measure changes in the chemical microenvironment of a biofilm during associated biophysical processes (Huang et al. 2018). Looking to the future, SECM is likely to become an increasingly important technology to investigate interspecies metabolic communication and growth competition, especially with additional advancements in our understanding of the oral metabolome.

Conclusion

The oral biofilm is the ideal biological niche to study molecular ecological questions of microbial community development, interspecies interactions, and biofilm biophysical properties. Nearly all organisms have been cataloged by oral microbiome sequencing efforts and a considerable number of species is cultivable. The development of genetic techniques has advanced tremendously (see the “Molecular Microbiology” section of this thematic issue) and allows for the targeted genetic manipulation of microbiome species important for oral biofilm ecology. Here, we describe important approaches that have played invaluable roles in our current understanding of oral biofilm ecology. With the ability to now model larger bacterial communities, we will hopefully begin to understand why specific microbial arrangements exist and how they develop to address many of the fundamental questions of microbial ecology that are critical for our understanding of oral and systemic health.

Conflict of interest statement

The authors declare no competing interests.