Polymicrobial Aggregates in Human Saliva Build the Oral Biofilm

Polymicrobial Aggregates in Human Saliva Build the Oral Biofilm Aurea Simon-Soro, a , b *Zhi Ren, a Bastiaan P. Krom, c Michel A. Hoogenkamp, c Pedro J. Cabello-Yeves, b §Scott G. Daniel, d Kyle Bittinger, d Inmaculada Tomas, e Hyun Koo, a , f Alex Mira b , g a Biofilm Research Laboratories, Department of Orthodontics, Divisions of Community Oral Health & Pediatric Dentistry, School of Dental Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA b Center for Advanced Research in Public Health, FISABIO Foundation, Valencia, Spain c Department of Preventive Dentistry, Academic Centre for Dentistry Amsterdam, University of Amsterdam and Vrije Universiteit Amsterdam, Amsterdam, The Netherlands d Division of Gastroenterology, Hepatology, and Nutrition, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA e Oral Sciences Research Group, Special Needs Unit, Department of Surgery and Medical-Surgical Specialties, School of Medicine and Dentistry, Universidade de Santiago de Compostela, Health Research Institute Foundation of Santiago (FIDIS), Santiago de Compostela, Spain f Center for Innovation & Precision Dentistry, School of Dental Medicine and School of Engineering & Applied Sciences, University of Pennsylvania, Philadelphia, Pennsylvania, USA g CIBER Center for Epidemiology and Public Health, Madrid, Spain Aurea Simon-Soro and Zhi Ren contributed equally. Author order was determined in order of decreasing seniority in the project. ABSTRACT Biofilm community development has been established as a sequential process starting from the attachment of single cells on a surface. However, microorganisms are often found as aggregates in the environment and in biological fluids. Here, we conduct a comprehensive analysis of the native structure and composition of aggregated microbial assemblages in human saliva and investigate their spatiotemporal attachment and biofilm community development. Using multiscale imaging, cell sorting, and computational approaches combined with sequencing analysis, a diverse mixture of aggregates varying in size, structure, and microbial composition, including bacteria associated with host epithelial cells, can be found in saliva in addition to a few single-cell forms. Phylogenetic analysis reveals a mixture of complex consortia of aerobes and anaerobes in which bacteria traditionally considered early and late colonizers are found mixed together. When individually tracked during colonization and biofilm initiation, aggregates rapidly proliferate and expand tridimensionally, modulating population growth, spatial organization, and community scaffolding. In contrast, most single cells remain static or are incorporated by actively growing aggregates. These results suggest an alternative biofilm development process whereby aggregates containing different species or associated with human cells collectively adhere to the surface as “growth nuclei”to build the biofilm and shape polymicrobial communities at various spatial and taxonomic scales. IMPORTANCE Microbes in biological fluids can be found as aggregates. How these multicellular structures bind to surfaces and initiate the biofilm life cycle remains understudied. Here, we investigate the structural organization of microbial aggregates in human saliva and their role in biofilm formation. We found diverse mixtures of aggregates with different sizes, structures, and compositions in addition to freeliving cells. When individually tracked during binding and growth on tooth-like surfaces, most aggregates developed into structured biofilm communities, whereas most single cells remained static or were engulfed by the growing aggregates. Our results reveal that preformed microbial consortia adhere as “buds of growth,”governing biofilm initiation without specific taxonomic order or cell-by-cell succession, which Editor Marvin Whiteley, Georgia Institute of Technology School of Biological Sciences Copyright © 2022 Simon-Soro and Ren et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license. Address correspondence to Alex Mira, [email protected], or Hyun Koo, [email protected]. *Present address: Aurea Simon-Soro, Department of Stomatology, School of Dentistry, University of Seville, Seville, Spain. §Present address: Pedro J. Cabello-Yeves, Evolutionary Genomics Group, Departamento de Producción Vegetal y Microbiología, Universidad Miguel Hernández, San Juan de Alicante, Alicante, Spain. The authors declare no conflict of interest. Received 21 January 2022 Accepted 24 January 2022 Published January/February 2022 Volume 13 Issue 1 e00131-22 ®mbio.asm.org 1 RESEARCH ARTICLE 22 February 2022 Downloaded from https://journals.asm.org/journal/mbio on 20 December 2022 by 150.214.182.238. provide new insights into spatial and population heterogeneity development in complex ecosystems. KEYWORDS polymicrobial aggregate, saliva, oral biofilm, spatial structure, microbiome Biofilms are highly organized microbial communities residing on surfaces (1–4). With few exceptions, biofilms harbor polymicrobial communities displaying complex spatial organization and interspecies interactions. Extensive efforts have been devoted to understanding the biofilm initiation and further development (5, 6). A key step is the initial attachment of microbial cells on a surface. Free-living bacteria (planktonic cells) can bind to the surface and start the formation of a matrix-enclosed microbial community, creating its own microenvironment, which varies depending on the host or other natural niches. The common model relies on a single species or a community of different species binding to the surface as single cells (7). However, microorganisms in natural settings or in biological fluids are often found forming multicellular aggregates in addition to planktonic cells (8–10). Bacteria in saliva or in synovial fluid usually form aggregates (11–13), while aggregated microbes are also found in aquatic ecosystems (14). Recent studies have assessed the role of aggregates in biofilm formation using individual single-species laboratory strains (8, 15). However, how naturally occurring aggregates modulate biofilm initiation and community development remains largely unexplored. Human saliva harbors a myriad of microbial species that interact and bind to several surfaces in the oral cavity, leading to the initiation of biofilms. Oral biofilms are one of the most diverse communities that have provided fundamental knowledge about biofilm biology and development (6). Microbes can contribute to oral and systemic health, but they also mediate oral diseases such as dental caries and periodontitis. Daily oral hygiene measures help to remove oral biofilmsfromthetoothsurface.However,oralbiofilms rapidly return, and the ecological succession model, whereby individual species colonize tooth surfaces in a sequential and orderly manner, has been widely demonstrated and adopted (1, 16–24). In this model, microbes are grouped into early colonizers (those that bind to the tooth surface within the first hours), which can provide attachment sites for other species, and late colonizers, which increase after days or weeks of biofilm development (25). Recent advances combining micron-scale biogeography and high-throughput sequencing enhanced the understanding of the spatial and phylogenetic scales of biofilm community arrangement (1, 26, 27). Analysis of dental biofilms (plaque) in their native state from clinical samples revealed a more complex structure and composition than previously considered, harboring spatially heterogeneous clusters where bacteria traditionally considered early colonizers occupy external positions in the biofilm (1, 3, 4), which is difficult to explain under the current model. Likewise, a diverse mixture of early and late colonizers can be found in biofilms in vivo (21, 22), suggesting a less ordered pattern of the initial colonizing bacteria. Although microbial aggregates were found in saliva decades ago, their structure, taxonomic organization, and roles in biofilm initiation and structuring have not been explored. Further elucidation on whether native-state aggregates in human saliva can influence biofilm initiation at spatiotemporal and phylogenetic scales may provide new fundamental insights. Here, we comprehensively analyze the composition and structure of the humanderived aggregated salivary community in its native state and their role in biofilm initiation using multiscale time-lapsed imaging, flow cytometry, and sequencing with computational analyses across different time points. We find naturally occurring aggregates with various sizes, structures, and compositions, including cobinding with human cells and free-living cells. In addition, early and late colonizers are found together in mixed-species aggregates. By tracking the entire population of surface colonizers, we discover that polymicrobial aggregates actively bind and rapidly proliferate, whereas single bacterial cells remain mostly static. Further analyses reveal that the aggregated consortia act as growth nuclei expanding across the surface, engulfing the static single cells to coordinate the Simon-Soro and Ren et al. ® January/February 2022 Volume 13 Issue 1 e00131-22 mbio.asm.org 2 Downloaded from https://journals.asm.org/journal/mbio on 20 December 2022 by 150.214.182.238. composition, size, and diversity of the biofilm community structure. Hence, our results suggest an addition to classical models of sequential single-cell binding and orderly ecological succession, whereby mixed microbes collectively adhere as buds of growth, leading to spatially and taxonomically heterogeneous communities, which may be applicable to other microbiomes harboring similar polymicrobial aggregates. RESULTS Native-state microbial structure in human saliva. We performed a comprehensive analysis of the structure and composition of the salivary microbial community in its native, intact form using a combination of super-resolution confocal imaging and scanning/transmission electron microscopy (SEM and TEM). We found that the microorganisms in human saliva rarely appeared in a single-cell free-living state. Rather, the microbes formed clusters, including some associated with desquamated oral epithelial cells (Fig. 1). The microbial aggregates in saliva ranged from 3 to 10 m mupto50 m m in diameter. We then used flow cytometry to investigate the size and complexity of the microbial structures in saliva from human subjects (20). Saliva samples were collected 30, 90, 150, and 360 min after toothbrushing. Fluorescence-activated cell sorting was employed to separate the different microbial constituents by sorting based on complexity (Fig. 2a). Data analysis revealed that, based on size, microbial structures in saliva clustered into 2 subpopulations, corresponding to a small fraction (,3 m m) and a large fraction (.3 m m). Microscopic examination indicated that the fraction under 3 m m contained mostly single cells. In contrast, we observed a highly heterogeneous population within the large fraction (.3 m m), including aggregates composed of different bacterial cells and FIG 1 Microbial community structure in human saliva in its native state. (a) Confocal microscopy of the structured microorganisms found in intact human saliva showing a highly complex mixture of single cells, microbial aggregates, and bacteria attached to host epithelial cells (arrowheads). (b) Scanning electron microscopy (b1) and transmission electron microscopy (b2) of polymicrobial aggregates found in human saliva. (c) Microorganisms in human saliva associated with desquamated oral epithelial cells as shown by SEM (c1) and TEM (c2) images. Salivary Microbial Aggregates Build Oral Biofilms ® January/February 2022 Volume 13 Issue 1 e00131-22 mbio.asm.org 3 Downloaded from https://journals.asm.org/journal/mbio on 20 December 2022 by 150.214.182.238. desquamated epithelial cells with bound microorganisms. The proportion of free-living cells was 3%, whereas 97% of microorganisms were found as aggregates based on cell counts in each fraction (see Table S1 in the supplemental material), indicating that the microorganisms in saliva were predominantly in aggregated form (versus planktonic single cells; P,0.001) (Fig. 2b). Bacterial composition within community structure in saliva. Next, we asked whether the salivary microbial compositions in single-cell and aggregated forms are different. Using 16S rRNA sequencing, we analyzed the bacterial composition in the small and large fractions sorted by flow cytometry. We observed that specific taxa can be found in aggregates such as Porphyromonas,Fusobacterium,andHaemophilus,whereas other species can be found as both free-living single cells and in aggregated communities, including Prevotella,Neisseria,Streptococcus,andVeillonella (Fig.2c).We then asked whether the different microbial species composition in the two sorted fractions was driven by the composition and distribution of the bacteria across all the subjects. We analyzed the beta-diversity using Jaccard distances and observed that samples clearly clustered by subject but not by community structure (Fig. 2d), indicating that both small and large fractions had similar bacterial compositions. Considering the salivary circadian rhythm might affect the structure and composition of the microorganisms, we also assessed both the small and large fractions at different time points. Our data suggested that the bacterial composition was variable between individuals and through time (Fig. 2e). Virtually all genera appeared to be present already at 30 min after toothbrushing (morning), including strict anaerobes traditionally considered late colonizers, like Prevotella or Veillonella. Genera classically considered early colonizers, like Streptococcus,Granulicatella,andGemella,were FIG 2 Bacterial composition within community structure in saliva. (a) Diagram of the workflow to assess and sort the microbial structures in saliva using flow cytometry. We sorted the sample into two subgroups based on the clustering of the flow cytometry data, described as small (,3 m m) and large (.3 m m) fractions. For cell counting, the large fraction was sonicated to disperse microbial aggregates and reanalyzed by flow cytometry. (b) Cell counts of the sorted cells (event numbers) in small and large fractions of saliva samples. (c) Sankey diagram showing bacterial composition abundance (line thickness) in each sorted fraction. (d) Principal correspondence analysis using Jaccard index analysis of the small and large (shapes) fractions of 8 subjects (color). (e) Bacterial abundance dynamics on each sorted fraction found through time of sample collection (0 to 360 min). ***,P,0.001 by Wilcoxon test. Simon-Soro and Ren et al. ® January/February 2022 Volume 13 Issue 1 e00131-22 mbio.asm.org 4 Downloaded from https://journals.asm.org/journal/mbio on 20 December 2022 by 150.214.182.238. found at higher levels after 6 h (afternoon). Interestingly, we observed some individuals showed a very constant pattern of bacterial composition in the different salivary size fractions and through time (PGN08), whereas others showed several bacterial genera preferentially found in aggregates or in the planktonic fraction (PGN09), with some temporal changes (Fig. S3a). Despite individual variability, the overall microbial distribution and composition were conserved, showing that in saliva (i) microbes are present primarily in the large (aggregate) fraction rather than in the small (single-cell) fraction, (ii) certain bacteria form primarily aggregates while others can be in both single-cell state and aggregated forms, and (iii) bacteria that traditionally have been considered late and early colonizers can be found mixed in the aggregate fraction. In addition, a preliminary analysis of the microbial composition of initially formed dental plaque samples (6 h posttoothbrushing) show that both early and late colonizers are all detected together (Fig. S3b), suggesting that they may bind concomitantly as mixed consortia to modulate biofilm formation. Hence, we explored whether single cells or aggregates preferentially attach to a hydroxyapatite surface to initiate and develop the biofilm. Seeding of saliva-derived biofilms. Biofilm formation involves bacterial colonization and further growth of the colonizers on a surface (6). The presence of aggregated microbial communities in saliva indicates that these structures are members of the initial colonizing community. To investigate whether aggregates in their native state form biofilms, we developed an ex vivo fluid-to-biofilm model using time-lapse microscopy within a microfluidics system combined with computational analyses (see Materials and Methods). We first characterized the microbial binding pattern (initial colonizers) on the hydroxyapatite surface (HA; a tooth enamel surrogate). Given that the microbial composition of saliva varies in different donors, we used pooled unstimulated saliva to assess the spatial distribution of initial microbial colonization to HA. The HA was incubated with the pooled saliva inoculum to allow bacterial binding before being aseptically transferred to a microfluidic imaging device. The HA surface was analyzed by confocal imaging of the colonized structures at various time points (30 to 240 min). By comparing the surface coverage rates, we first determined that maximum binding was reached after 60 min of incubation (Fig. S1). We then assessed the composition of microbial species in the fluid phase (saliva inoculum) and solid phase (surface bound). Notably, we found that a diverse mixture of traditionally described early and late colonizers (17–20, 26) in saliva could bind simultaneously on the HA surface (Fig. 3a and Fig. S4). Early colonizers detected in our model include Streptococcus,Haemophilus, Capnocytophaga,Actinomyces,andVeillonella, whereas late colonizers, such as Prevotella, Treponema,Aggregatibacter,andPorphyromonas, were also attached on the apatitic surface, suggesting mixed-species cocolonization during the binding process. Fusobacterium nucleatum, a previously described bridging organism, was also found in this initial colonizing community. The entire population of surface colonizers was then imaged using our optimized biofilm imaging platform (4, 26) at multiple length scales. We foundalargepopulationofsurface-boundbacteria(n.50 in an area of 319.45 by 319.45 m m 2 ), comprised of a diverse mixture of single cells, bacterial chains, and highly heterogeneous microbial aggregates, as initial colonizing units (Fig. 3b1 and b2). We employed BiofilmQ (28) to characterize the individual structures of the colonizing community. This was performed by segmenting the image subsets containing individual units at the regions where the bacteria attached after the 60 min binding period (t 0 ). The biovolumes of individual units varied from ;2to250 m m 3 (Fig. 3c). We further grouped all the initial colonizing units based on their biovolume (V): small clusters with Vof #10 m m 3 were definedassinglecells(26),and large clusters with Vof .10 m m 3 were defined as aggregates (Fig. 3c). The data show a remarkably diverse colonizing community (or colonizing units) comprised not only of single cells but also aggregates of various sizes and shapes bound on Salivary Microbial Aggregates Build Oral Biofilms ® January/February 2022 Volume 13 Issue 1 e00131-22 mbio.asm.org 5 Downloaded from https://journals.asm.org/journal/mbio on 20 December 2022 by 150.214.182.238. the apatitic surface. The initial colonizing units may provide distinctive starting points for further biofilm growth. Dynamics of biofilm initiation and growth from saliva fluid. Considering that both aggregates and single cells are present in saliva and bound to apatitic surfaces, we investigated how this colonizing community mediates biofilm initiation spatiotemporally. To investigate the growth dynamics of surface-attached saliva-derived microbes, we applied a high-resolution confocal imaging system coupled with flow-cell microfluidics that was optimized for biofilm time-lapsed imaging (Fig. 4a) (26). We used filter-sterilized pooled saliva as the culture medium for the fluid-to-biofilm model to mimic the natural nutrient condition in the oral cavity. Upon initiating the flow in the microfluidic device, biofilm growth was observed across the HA, and image acquisition was performed every 30 min to generate a 4-dimensional data set (x,y,z, and time frame) for further analysis (Fig. 4b, top, and Movie S1). We found that a subset of the initial colonizing units actively grew and evolved into larger biofilm structures (Movie S1). To investigate whether the initial structure had an impact on the growth dynamics, we applied geometrical filtering to assign all the initial colonizing units into two subgroups based on their initial biovolume (Vt 0 ) as single cells (Vt 0 #10 m m 3 ) or aggregates (Vt 0 .10 m m 3 ). Because bacterial cells can detach under flow and cause discrepancy in the calculation of the growth dynamics, we only analyzed the individual colonizing units that were attached at the initial binding and remained bound on the surface throughout the experiment (until 360 min). Each colonizing unit was monitored individually using the cube tracking algorithm in BiofilmQ. Biofilm structures originating from the same starting point (i.e., the same colonizing unit) were assigned the same track identifier (ID; color coded) so that each unit could be traced individually FIG 3 Seeding of saliva-derived biofilms. (a) Microbial species found in the fluid phase (saliva inoculum) and solid phase (binding). Low-frequency bacteria are represented as “Other”and include Treponema denticola,Aggregatibacter actinomycetencomitans,Tannerella forsythia, and Porphyromonas gingivalis. Early colonizers (e.g., Streptococcus,Haemophilus,Capnocythophaga,Actinomyces, and Veillonella) together with species traditionally considered late colonizers (e.g., Prevotella,Porphyromonas, and Treponema) were found on the surface after the initial binding, suggesting mixed-species cocolonization during the binding process. Fusobacterium nucleatum, a bridging organism, was also found in this initial colonizing community. Data represent pooled samples from three individuals, and three independent experiments were performed. (b) Representative image (Z projection) of in situ visualization of surface colonizing units on the saliva-coated tooth-mimetic hydroxyapatite surface. (b1) A highly diverse and heterogenous structural organization, including both single cells/clusters (in white box) and large aggregates (in cyan box), can bind to the surface. (b2) Magnified images of the single cells/ small clusters (as shown in panel b1). (c) Initial biovolume of the individual colonizing units was analyzed using BiofilmQ software based on the 3- dimensional confocal data sets. A threshold biovolume (V)of10 m m 3 (black dotted line) was used to group the initial colonizing units: small clusters with V of #10 m m 3 were defined as single cells (depicted in pink), and large clusters with Vof .10 m m 3 were defined as aggregates (blue). The results represent a large population of initial colonizing units (N.300). Imaging and metagenomics data were obtained from at least 3 independent experiments. Simon-Soro and Ren et al. ® January/February 2022 Volume 13 Issue 1 e00131-22 mbio.asm.org 6 Downloaded from https://journals.asm.org/journal/mbio on 20 December 2022 by 150.214.182.238. from the initial binding (t 0min ) (Fig. 4b, bottom, and Movie S1). We found that most aggregates steadily grow, whereas most single-cell units stay static (yellow arrow in Fig. 4b) or are engulfed by a growing aggregate (white arrow). To further visualize the spatial growth distribution of the colonizing units, we separated those growing from either single cells (depicted in pink) or aggregates (in blue), as shown in Fig. 4c. The data suggest that the biofilm community growth on the surface originated mostly FIG 4 Dynamics of biofilm initiation and growth from saliva fluid. (a) Experimental setup for fluid-to-biofilm model used for assessing the colonization of microorganisms in saliva fluid (t= 0 min) and further growth under cell-free saliva flow in real time. (b, upper) Time-lapsed confocal images (represented as Z projection and orthogonal projection) of biofilm initiation from saliva fluid. (Lower) Three-dimensional time-lapsed tracking map of biofilm growth. Colors indicate individual biofilm clusters that originated from different initial colonizing units. Only the clusters that attached at the initial binding (t= 0 min) and remained bound until the end of the experiment (t= 360 min) were tracked. Yellow arrow, a colonized single cell that stayed static; white arrow, attached single cells that remained static and were engulfed by a growing aggregate. (c) The spatial distribution of biofilm growth by origin (single cell or aggregate). Blue, biovolume originated from aggregates; pink, biovolume originated from single cells. (d) Growth curves of individual colonizing units within 360 min after initial colonization. (Left) Most single cells did not grow or grew slowly despite remaining attached to the surface (depicted in pink). (Right) Most initial bound aggregates grew steadily, with some reaching large sizes (depicted in blue). (e) Fold change of single cell or aggregate growth. Salivary Microbial Aggregates Build Oral Biofilms ® January/February 2022 Volume 13 Issue 1 e00131-22 mbio.asm.org 7 Downloaded from https://journals.asm.org/journal/mbio on 20 December 2022 by 150.214.182.238. from aggregates rather than from single cells. In addition, we also compared the total biovolume (biomass) of all tracked colonizing units from the beginning (t 0min ) to the end of the experiment (t 360min ) on an area of 159.73 by 159.73 m m 2 . As shown in Fig. S5, the total biovolume from an aggregate origin increased over time, accounting for .90% of the total amount at the endpoint (;17,000 m m 3 ). In contrast, single cellderived biovolume remained low, resulting in less than 1,500 m m 3 at 360 min. Next, we further analyzed the biovolume-based growth curves of each colonizing unit and compared its dynamic changes. Our data revealed that most initial colonizing units grouped as single cells did not grow or grew slowly despite remaining attached to the surface throughout the experiment period (Fig. 4d, left, and e). Only 42% of single cells increased in biovolume with a median fold change of 1.83-fold (Fig. 4e). In contrast, most initial colonizing units bound as aggregates grew steadily (79% of aggregates increased with a median fold change of 11.2-fold; P=0.0096), reaching large sizes ranging between 2,000 and 6,000 m m 3 at 360 min (Fig. 4d, right; Fig. 4e). Thus, two distinctive phenotypes were observed: (i) static or slow-growing colonizing units, predominantly single cells, whereby the bacteria bound to the HA maintained their biovolume relatively constant or with limited changes, and (ii) active-growing units, predominantly comprised of aggregates of various sizes that grew faster and expanded their biovolume over time, often engulfing the static population. Given the contrasting growth behavior between aggregated consortia and single cells, we hypothesized that the aggregated form could act as biofilm initiators promoting the spatial growth and structural heterogeneity of the microbial community. Using the same model for growth dynamics tracking, we compared the biofilm initiation by saliva inoculum from the same pooled source but processed in two different ways: (i) the native inoculum, which contained intact communities of single cells and aggregates, and (ii) the dispersed inoculum, which was sonicated to disperse the natural aggregates without impacting bacterial viability. Our results showed that colonizing units in the dispersed group, comprised of mostly single cells and smaller aggregates, grew minimally with most of the cells remaining static, whereas the native community showed more active growth forming enlarged structures (Fig. 5a1). Furthermore, the biofilm initiating from the native inoculum displayed highly heterogeneous features, including peak-like architectures and a diverse range of thicknesses (Fig. 5a2, top, and b). In sharp contrast, the biofilm originating from the dispersed inoculum showed smaller and thinner structures that were like those in the dispersed colonizing community (Fig. 5a2, bottom, and b). Moreover, we found that bacteria naturally attached to human epithelial cells (host cell-microbial aggregates) can also grow faster than bacteria alone and proliferate rapidly to form large biofilm structures (Movie S2, Fig. S6). Hence, our results suggest that aggregated bacteria in the native state act as growth nuclei and play a pivotal role in the initiation and development of spatial heterogeneity of the biofilm community. Taxonomic scale of native and dispersed communities. Saliva, as a source for the biofilm formation, harbors hundreds of bacterial taxa (29). Here, we analyzed the microbial composition in samples from the native and dispersed communities, as shown in Fig. 5a. We collected different samples in a sequential manner: inoculum (baseline), binding (initial colonizing units), and 6-h biofilm. Samples for quality control and reproducibility check were also included (Fig. S2). We found that dispersion of the aggregates did not influence the types of bacteria that were attached to the HA surface, showing alpha and beta diversity differences were related to different time points but not between native and dispersed states (P,0.001) (Fig. 5c). Furthermore, we analyzed the relative abundance of bacterial genera at the three stages (inoculum, binding, and biofilm) (Fig. 5d). Notably, we detected 155 bacterial taxa bound to the HA from the 203 ASV (16S amplicon sequence variant) found in human saliva inoculum, showing that saliva richness is represented during biofilm initiation. After initial biofilm growth, the microbiota was reduced to 53 taxa, which was enriched by oxygen-toler- ant bacteria compatible with the aerobic environment dominating biofilm initiation Simon-Soro and Ren et al. ® January/February 2022 Volume 13 Issue 1 e00131-22 mbio.asm.org 8 Downloaded from https://journals.asm.org/journal/mbio on 20 December 2022 by 150.214.182.238. under our experimental conditions. Altogether, we found a highly diverse colonizing community, suggesting that microbial attachment is not exclusive to single cells or specific species but also encompasses a wide range of microbial aggregates containing a mixture of different species. Neither sequential nor orderly succession of selective binding of specific species was observed. Rather, the aggregates bound on the surface proliferate rapidly and expand tridimensionally, modulating population growth dynamics, spatial organization, and community scaffolding. DISCUSSION Our findings reveal how naturally formed aggregates in the human oral environment mediate biofilm formation at different spatial, temporal, and phylogenetic scales. We show four key features, as summarized in Fig. 6. (i) Saliva, the source of microbes, across different subjects and timescales is dominated by polymicrobial aggregates harboring different species together in addition to free-living bacterial cells. Notably, bacteria are also found coadhered with human epithelial cells, indicating a more complex and structurally diverse community than previously described (7). (ii) Bacteria that traditionally have been considered early and late colonizers are found together in the adhering aggregates and in early formed human plaque samples, indicating that they bind concomitantly to the surface as a colonizing unit. (iii) Microbial aggregates bound FIG 5 Spatial and taxonomic scale of native and dispersed communities. (a) Biofilm initiation from the native-state saliva inoculum versus the dispersed saliva inoculum. (a1) Confocal images (represented as Z projection and orthogonal projection) of biofilm originating from the native state (top) and the dispersed inoculum (bottom). (a2) Three-dimensional rendering of the in situ biofilm morphological analysis showing the 0-min and 360-min biofilm structures. (b) Results of biofilm thickness mapping showing the local thickness distribution at 360 min at four different cross sections (as shown in panel a2). (c1) Principal correspondence analysis using weighted UniFrac distances of the native and dispersed groups of human saliva used as inoculum and the biofilm initiation stages (binding and biofilm at 6 h). (c2) Richness analysis of the biofilm stages. Median is indicated in each group with the black symbol. P,0.001 by pairwise comparisons using Kruskal-Wallis H test. (d) Bacterial relative abundances of the main genera of native and dispersed groups. Salivary Microbial Aggregates Build Oral Biofilms ® January/February 2022 Volume 13 Issue 1 e00131-22 mbio.asm.org 9 Downloaded from https://journals.asm.org/journal/mbio on 20 December 2022 by 150.214.182.238.