An in vitro biofilm model system maintaining a highly reproducible species and metabolic diversity approaching that of the human oral microbiome

Abstract

Background: Our knowledge of microbial diversity in the human oral cavity has vastly expanded during the last two decades of research. However, much of what is known about the behavior of oral species to date derives from pure culture approaches and the studies combining several cultivated species, which likely does not fully reflect their function in complex microbial communities. It has been shown in studies with a limited number of cultivated species that early oral biofilm development occurs in a successional manner and that continuous low pH can lead to an enrichment of aciduric species. Observations that in vitro grown plaque biofilm microcosms can maintain similar pH profiles in response to carbohydrate addition as plaque in vivo suggests a complex microbial community can be established in the laboratory. In light of this, our primary goal was to develop a robust in vitro biofilm-model system from a pooled saliva inoculum in order to study the stability, reproducibility, and development of the oral microbiome, and its dynamic response to environmental changes from the community to the molecular level.

Results: Comparative metagenomic analyses confirmed a high similarity of metabolic potential in biofilms to recently available oral metagenomes from healthy subjects as part of the Human Microbiome Project. A time-series metagenomic analysis of the taxonomic community composition in biofilms revealed that the proportions of major species at 3 hours of growth are maintained during 48 hours of biofilm development. By employing deep pyrosequencing of the 16S rRNA gene to investigate this biofilm model with regards to bacterial taxonomic diversity, we show a high reproducibility of the taxonomic carriage and proportions between: 1) individual biofilm samples; 2) biofilm batches grown at different dates; 3) DNA extraction techniques and 4) research laboratories.

Conclusions: Our study demonstrates that we now have the capability to grow stable oral microbial in vitro biofilms containing more than one hundred operational taxonomic units (OTU) which represent 60-80% of the original inoculum OTU richness. Previously uncultivated Human Oral Taxa (HOT) were identified in the biofilms and contributed to approximately one-third of the totally captured 16S rRNA gene diversity. To our knowledge, this represents the highest oral bacterial diversity reported for an in vitro model system so far. This robust model will help investigate currently uncultivated species and the known virulence properties for many oral pathogens not solely restricted to pure culture systems, but within multi-species biofilms.

Figure 1

Correspondence analysis showing reproducibility and 16S profile similarities within biofilm and saliva samples. Axis 1 explains 47% of the variation in the dataset; axis 2 explains 19% of the variation. Replicate biofilm samples representing batches 1 and 3 cluster closely together whereas batch-2 biofilms that derive from an SIP experiment cluster more distantly along the first ordination axis. Saliva-derived replicates (1 to 3) also show similar 16S diversity.

Figure 2

Phylogenetic tree based on 1,642 Human Oral Microbiome Database (HOMD) reference sequences. 16S genes that were obtained from pyrosequencing in this study were matched based on sequence homology with HOMD reference sequences. Red and blue bars indicate normalized relative abundance counts of human oral taxon (HOT) designations that were identified in the saliva and biofilm samples, respectively. Sequence matches between reference sequences and query sequences were counted at 97% sequence identity cutoff and 95% sequence coverage.

Figure 3

Bubble charts showing 16S profile reproducibility at human oral taxon (HOT) designation level (98.5% sequence similarity) in replicate biofilm growth wells. Batches 2 and 3 were grown at different time points and batch 2 was also subjected to an additional centrifugation step during DNA extraction. Cultivated species, both unnamed and named batches, according to HOMD classification are presented in A. Uncultured phylotypes are presented in B. Read-abundances were log transformed and all identified HOTs are presented here. Bubble sizes correspond to relative abundance values calculated from log-transformed 16S read-abundance data for each HOT. Standard deviation in HOT abundance between wells from the same batch ranged between 0.1% and 3% whereas standard deviation was higher (0.3 to 8%) between batches.

Figure 4

Metabolic profile comparisons of Human Microbiome Project (HMP) oral metagenomes, other HMP-body sites and the in vitro biofilm metagenomes. The METAREP tool was used to compare Kyoto Encyclopedia of Genes and Genomes (KEGG) super pathways representing the 12- and 16-hrs-old in vitro biofilms from this study with HMP-oral metagenomes representing supragingiva, keratinized gingiva and saliva samples from healthy subjects. These oral metagenomes were also compared to HMP metagenomes from other body sites, that is, anterior nares (external position of nostrils), mid vagina and posterior fornix (area behind lower portion of uterus) by using hierarchical cluster analyses of ORF abundance data for each sample. Colored bar in the top indicates relative abundance in percentage of annotated ORFs that fall within each KEGG super pathway (blue, approximately 0%; white, approximately 6%; red, approximately 16%).

Figure 5

Temporal metagenomic read-abundance profiles from MetaPhlan analyses of bacterial species. Relative abundance of paired-end reads that were classified at bacterial species-level based on clade-specific marker genes from 3,000 reference genomes [47]. pH at the different time points is shown in white text inside the bar graph. No pH measurement was available for 48 hrs (N/A). Numbers of estimated bacterial species from MetaPhlan analyses are shown inside the bar graph (No. sp.).