Oral Biofilms Rely on Architecture, Not Just Which Microbes Are Present

The bacteria that live on our teeth do not behave like the same microbes floating freely in saliva. A wide-ranging review published in Frontiers in Cellular and Infection Microbiology argues that oral biofilms, the structured microbial communities behind tooth decay, gum disease and infections around dental implants, are governed by spatial organization, extracellular matrix construction and metabolic exchange as much as by which species are present. The authors, Abdulrahman Abdullah Almusallam and Khaled S. Allemailem of Qassim University in Saudi Arabia, synthesize molecular and ecological evidence to make a pointed argument: reducing biofilm biomass is not the same as restoring a healthy community, and many anti-biofilm strategies judged by killing alone may miss what actually drives disease.

Imaging of human dental plaque has revealed reproducible micron-scale arrangements in which particular taxa occupy preferred positions relative to one another rather than being distributed randomly. That positioning carries functional weight, because cells only a few micrometres apart can face markedly different nutrient concentrations, oxygen levels, pH values and exposure to inhibitory metabolites. In cariogenic plaque, structured assemblies containing Streptococcus mutans can generate localized acidic zones at the tooth surface, linking microscale organization to an environment relevant to enamel demineralization. Where activity concentrates within the biofilm, in other words, is itself a product of the community’s architecture.

The review traces how these communities assemble. Enamel is rapidly covered by an acquired pellicle derived from salivary and gingival components, so arriving bacteria encounter adsorbed host proteins rather than bare mineral. Streptococcus gordonii uses its antigen I/II-family proteins SspA and SspB, together with the serine-rich adhesin Hsa, to bind immobilized salivary agglutinin, a redundancy that may help pioneer colonizers stay attached as pellicle composition varies. Genetic dissection of Actinomyces oris type-2 fimbriae pinned coaggregation with Streptococcus oralis down to a single tip protein, CafA: deleting the gene abolished the interaction, and both purified CafA and antibodies against it blocked it. Established cells then become receptors for newcomers. The minor fimbrial protein Mfa1 of Porphyromonas gingivalis recognizes a region of the S. gordonii protein SspB, a contact needed for efficient P. gingivalis biofilm development on streptococcal substrata, while the bridging organism Fusobacterium nucleatum relies on the outer-membrane adhesin RadD, whose expression is controlled by the CarRS two-component system and inhibited by free lysine.

Signaling adds a regulatory layer to these physical contacts. Early genetic work linked the ComCDE competence system of S. mutans to biofilm architecture, and sender-responder experiments showed that ComS/XIP-dependent signals pass between cells, including within biofilms. But the response depends heavily on context: mature biofilms need more signal to respond, and activation of the competence gene comX becomes increasingly heterogeneous, with only part of the population reacting at all. Autoinducer-2 supports mutualistic growth in some saliva-fed dual-species communities, yet the review cautions that a luxS deletion phenotype alone does not prove a dedicated quorum-sensing role, since LuxS also feeds the activated methyl cycle. A 2026 study found detectable AI-2 production limited to Fusobacterium animalis subsp. among strains examined, undermining assumptions of a conserved AI-2 system across the F. nucleatum species complex.

The extracellular matrix emerges as a dynamic continuum with the cell surface rather than inert packing material. Glucosyltransferases GtfB, GtfC and GtfD of S. mutans produce different glucans, and losing GtfB or GtfC changes matrix accumulation and three-dimensional structure, arguing for complementary rather than redundant roles. Glucan-binding proteins GbpA, GbpC and GbpD each shape biofilm architecture differently, showing that matrix synthesis and matrix recognition are separate processes. Extracellular DNA arises through both lysis-associated and membrane-vesicle-associated routes, and competence-stimulating peptide can induce localized death in a subset of LytF-expressing cells that releases nucleic acids near the biofilm base. Even amyloid formation, once imagined as an early scaffold, appears temporally restricted: subsequent work found amyloid concentrated mainly in the nonadherent, detached fraction of aging S. mutans biofilms, with the adhesin P1 losing its adhesive function once amyloidogenesis occurred.

Metabolism within these communities is intensely local. Confocal pH ratiometry in in-situ-grown human dental biofilms detected differences approaching two pH units among regions within the same biofilm, with pH tending to be lower at the base than at the upper surface after carbohydrate challenge. Microelectrode measurements showed rapid oxygen consumption near the surface and effectively anoxic deeper regions, and new pH-oxygen ratiometry indicates that low pH and low oxygen can co-occur, though their relationship depends on the community, correlating more clearly in complex salivary biofilms than in monospecies S. mutans biofilms. Cross-feeding illustrates the interdependence: Veillonella atypica induced S. gordonii amylase expression only in nearby mixed microcolonies under flow, and a 2024 five-species model showed that high glucose plus high lactate drove rapid acidification and S. mutans outgrowth, while low glucose with abundant lactate favored Veillonella parvula and kept the bulk environment less acidic.

Host factors shape the community in ways that go beyond killing. Purified salivary MUC5B and its O-glycans suppress S. mutans competence signaling and cut genetic transformation without measurably slowing growth, while health-associated organisms such as S. gordonii and Actinomyces naeslundii draw nutrients from the same molecule. Histatin-5 protects oral tissue against Candida albicans in an ex vivo model, but a 2023 analysis found no comparable killing of seven tested oral streptococci at physiologically relevant concentrations, making the common description of histatin-5 as a broad-spectrum oral antimicrobial inaccurate. In periodontitis, a murine model showed that low-abundance P. gingivalis altered the resident microbiota and produced complement-dependent bone loss, with C5aR-TLR2 crosstalk blunting phagocytic clearance while leaving inflammatory signaling intact. The authors stress this keystone-like behavior does not imply P. gingivalis alone accounts for human periodontitis.

The review is equally skeptical about computational shortcuts. Machine learning can identify multivariate disease signatures, and a longitudinal early-childhood-caries cohort showed microbiome composition predicted subsequent disease with moderate discrimination before clinical onset. But a 2026 periodontitis classifier that achieved very high internal performance showed substantially variable results across three external datasets, leading the authors to conclude that internal AUC alone is a poor measure of clinical transportability. SHAP feature-importance values explain why a fitted model made a prediction; they do not demonstrate that changing the featured organism will change disease. Genome-scale metabolic models, including the iSMU model of S. mutans, whose hypotheses about sorbitol and complex-carbohydrate metabolism were validated experimentally, are praised when their predictions are converted into falsifiable experiments, but model-predicted metabolite exchanges remain predictions until confirmed by isotope tracing, supplementation or pathway disruption.

Emerging interventions are graded by evidence tier, and the grades are sobering. A dentifrice containing an antigen I/II-derived peptide reduced plaque mutans streptococci in a human study, proof-of-mechanism but no demonstrated caries-incidence benefit. The STAMP peptide C16G2 reduced plaque and salivary S. mutans in a pilot human intra-oral model while total bacterial levels changed relatively little. Catalytic ferumoxytol and dextran-coated iron oxide nanozymes exploit acidic microenvironments and reduced caries severity in animal models, but human long-term safety and ecological effects remain unknown. Probiotics show strain- and indication-dependent efficacy: topical Streptococcus dentisani altered salivary parameters in a randomized study without testing caries incidence, while a 2026 randomized trial of Limosilactobacillus reuteri as a periodontal adjunct in patients with diabetes found no significant additional clinical or microbiological benefit. A 2024 Cochrane review of 50 randomized trials found very-low-certainty evidence for clinically important six-month benefit of antimicrobial photodynamic therapy.

The authors’ central translational warning is that biofilm control and disease control are different goals. Engineered salivary peptides reduced enamel demineralization without significantly reducing S. mutans biofilm biomass, an effect that would have been missed had biomass been the sole endpoint. Targeted killing does not automatically restore a healthy microbiome: after S. mutans depletion, C16G2 also changed the structure of the surrounding community, and periodontal communities may remain transcriptionally distinct from health even after clinical improvement. Donor variability compounds the problem, with subgingival microcosm biofilms clustering strongly by inoculum source and periodontal status. The proposed path forward is staged perturbation, from monospecies mechanism through defined multispecies interaction, patient-derived microcosm and host-associated systems to human testing, with recovery phases after treatment withdrawal to distinguish genuine ecological reprogramming from temporary suppression. The unresolved challenge, the review concludes, is not finding stronger antibiofilm agents but determining which biofilm property must change, in which patients, for long enough, to improve oral health.

Subject of Research: Molecular architecture, host modulation and precision anti-biofilm strategies in multispecies oral biofilms

Article Title: Decoding multispecies oral biofilms: molecular architecture, host modulation and systems-guided precision anti-biofilm strategies

Article References: Almusallam, A. A., & Allemailem, K. S. (2026). Decoding multispecies oral biofilms: molecular architecture, host modulation and systems-guided precision anti-biofilm strategies. Frontiers in Cellular and Infection Microbiology, 16, Article 2002174. https://doi.org/10.3389/fcimb.2026.2002174

Image Credits: AI Generated

DOI: 10.3389/fcimb.2026.2002174

Keywords: oral biofilms, dental plaque, extracellular matrix, Streptococcus mutans, coaggregation, quorum sensing, metabolic cross-feeding, functional genomics, machine learning, precision therapy, periodontitis, dental caries