Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Mar 15, 2011; 108(Suppl 1): 4592–4598.
Published online Oct 11, 2010. doi:  10.1073/pnas.1011383107
PMCID: PMC3063583
Colloquium Paper

Human oral, gut, and plaque microbiota in patients with atherosclerosis


Periodontal disease has been associated with atherosclerosis, suggesting that bacteria from the oral cavity may contribute to the development of atherosclerosis and cardiovascular disease. Furthermore, the gut microbiota may affect obesity, which is associated with atherosclerosis. Using qPCR, we show that bacterial DNA was present in the atherosclerotic plaque and that the amount of DNA correlated with the amount of leukocytes in the atherosclerotic plaque. To investigate the microbial composition of atherosclerotic plaques and test the hypothesis that the oral or gut microbiota may contribute to atherosclerosis in humans, we used 454 pyrosequencing of 16S rRNA genes to survey the bacterial diversity of atherosclerotic plaque, oral, and gut samples of 15 patients with atherosclerosis, and oral and gut samples of healthy controls. We identified Chryseomonas in all atherosclerotic plaque samples, and Veillonella and Streptococcus in the majority. Interestingly, the combined abundances of Veillonella and Streptococcus in atherosclerotic plaques correlated with their abundance in the oral cavity. Moreover, several additional bacterial phylotypes were common to the atherosclerotic plaque and oral or gut samples within the same individual. Interestingly, several bacterial taxa in the oral cavity and the gut correlated with plasma cholesterol levels. Taken together, our findings suggest that bacteria from the oral cavity, and perhaps even the gut, may correlate with disease markers of atherosclerosis.

The human body is home to microbial ecosystems (microbiotas) whose structure and function differ between different sites in the body (16). These microorganisms outnumber the number of eukaryotic cells in the human body by at least an order of magnitude (7). The gut microbiota is the best-studied human-associated ecosystem and has a major impact on our physiology, immune system, and metabolism. For instance, obese individuals generally have a less diverse gut microbiota, and some studies have observed reduced levels of the bacterial phyla Bacteroidetes (3), although others have not (8, 9). Furthermore, germ-free mice have reduced adiposity and are resistant to diet-induced obesity (10, 11). Thus the gut microbiota can be considered an environmental factor that affects obesity. However, the role of the human microbiome in obesity-related metabolic diseases such as atherosclerosis remains to be explored.

Atherosclerotic disease, with manifestations such as myocardial infarction and stroke, is the major cause of severe disease and death among subjects with obesity. The disease is characterized by accumulation of cholesterol and recruitment of macrophages to the arterial wall. It can thus be considered both a metabolic and an inflammatory disease (12). Since the first half of the 19th century, infections have been thought to cause or promote atherosclerosis by augmenting proatherosclerotic changes in vascular cells (13). These changes include increased scavenger receptor expression and activity, enhanced uptake of cholesterol and modified LDL, increased expression of adhesion molecules and inflammatory cytokines, and other effects, such as stimulating macrophages to express cytokines, leading to atherosclerotic plaque vulnerability (13).

Epidemiological studies support an association between cardiovascular disease and infections, such as periodontal disease and Chlamydia infections (14, 15). Dental disease has been associated with elevated risk of myocardial infarction (14), and metabolic activity of the gut microbiota was recently shown to relate to blood pressure (16). Furthermore, in a study where bacterial DNA was identified in atherosclerotic plaques, 51.5% of the patients tested positive for Chlamydia in their atheromas (17). Several studies suggest an oral source for atherosclerotic plaque-associated bacteria (1821). However, to date, no single study has directly compared the microbial diversity of oral, gut, and atherosclerotic plaque microbiotas within individuals. This type of cross-site comparison is essential given the high level of variability observed in the microbiota between different subjects (16).

Here, we characterized the atherosclerotic plaque, oral, and gut microbiotas obtained from patients with atherosclerosis and healthy controls by pyrosequencing their 16S rRNA genes. Our study addressed the following questions: Is there a core atherosclerotic plaque microbiota? Are bacteria present in the plaque also detectable in the oral cavities or guts of the same individuals? Do the microbiotas of the oral cavity, gut, and atherosclerotic plaque relate to disease markers such as plasma levels of apolipoproteins and cholesterol? Is an altered oral or fecal microbiota associated with atherosclerosis? Our findings revealed a number of phylotypes common to the atherosclerotic plaque and oral and gut samples within individuals, and that the abundances of specific members of the oral and gut microbiota correlated with disease biomarkers.


Overall Comparison of the Human Oral, Gut, and Atherosclerotic Plaque Microbiotas.

We surveyed the atherosclerotic plaque, oral cavity (swab from periodontium area), and gut (feces) bacterial communities of 15 patients with clinical atherosclerosis and 15 age- and sex-matched healthy controls (Table 1). The 5′ variable regions (V1–V2) of the bacterial 16S ribosomal RNA (rRNA) gene were PCR amplified using barcoded primers 27F and 338R (22). We generated a dataset of 380,501 high-quality 16S rRNA sequences (n = 73: 15 patient and control fecal samples, 14 patient oral samples, 15 control oral samples, and 14 atherosclerotic plaque samples). Sequences were assigned to species-level operational taxonomic units (OTUs) using a 97% pairwise-identity cutoff, and chimera checking revealed that 3.1% of total sequences were putative chimeras. One atherosclerotic plaque sample was excluded from the downstream analysis due to low sequence counts (<1,700 sequences). The final dataset included representatives of 13 bacterial phyla; the majority of the sequences were classified as Firmicutes (63.8%), Bacteroidetes (11.7%), Proteobacteria (15.4%), and Actinobacteria (6.4%).

Table 1.
Characteristics of study participants

We compared the overall bacterial community composition using the unweighted UniFrac distance metric, a phylogenetic tree-based metric ranging from 0 (distance between identical communities) to 1 (distance between totally different communities with no shared ancestry). This analysis revealed strong clustering of samples by body site, with the atherosclerotic plaque samples forming a distinct cluster apart from oral and fecal samples, indicating all three sites have distinct microbial communities (Fig. 1 and Fig. S1). We used the unweighted (qualitative) rather than the weighted (quantitative) version of the metric because it generally performs better for resolving human body sites (4). The average phylogenetic diversity (23) of the microbiotas was similar for atherosclerotic plaque (AP) and oral cavity (OC, ratio AP/OC= 1.09), but highest for the gut (G; AP/G = 0.7, OC/G = 0.64).

Fig. 1.
Bacterial diversity clustering by body habitat. The first two principal coordinates (PC1 and PC2) from the principal coordinate analysis of unweighted UniFrac are plotted for each sample. Each symbol represents a sample, colored by body habitat (green, ...

Characterization of the Atherosclerotic Plaque Microbiota.

The atherosclerotic plaque contained low but detectable amounts of bacterial DNA: qPCR analysis revealed a positive correlation (ρ = 0.68, P = 0.009) between the amount of bacterial 16S rDNA and the number of leukocytes within the atherosclerotic plaques, suggesting that the amount of bacteria contributes to the inflammatory status of the atherosclerotic plaque. Compared with the oral and gut samples, the atherosclerotic plaque contained significantly higher levels of Proteobacteria and fewer Firmicutes (Fig. 2 A–E). We detected several OTUs present in all atherosclerotic plaque samples, and that differentiated these samples from oral and fecal samples. For instance, Chryseomonas (recently reclassified as Pseudomonas luteola), was detected at high levels in the atherosclerotic plaque samples and not at all in the gut or oral samples; furthermore, nearest shrunken centroids (NSC) analysis revealed that this OTU was the most discriminant genus between sites (i.e., driving the differences between the microbiotas of the different body sites; Fig. 3 A and B). In addition, three OTUs belonging to the genus Staphylococcus, three OTUs classified as Propionibacterineae, and one OTU belonging to the genus Burkholderia (Fig. 3 A and B) were specific for atherosclerotic plaques and present in all samples. Together, these observations support the notion that a “core” microbiota composed of the same genus-level lineages exists in atherosclerotic plaque.

Fig. 2.
Mean phylum abundances by body habitat for patients and controls. Plotted values are mean sequence abundances in each phylum for 1,700 randomly selected sequences per sample. (A) Atherosclerotic plaque; (B) oral cavity, patient; (C) oral cavity, control; ...
Fig. 3.
Members of the microbiota that differ in abundance between the body sites. (A) Shrunken differences for the 10 genera accounting for the differences among the three body sites. For each genus listed in center, the direction of the horizontal bars indicates ...

Composition of the Oral Microbiota and Its Relation to Atherosclerosis and Disease Markers.

The oral microbiota of patients and healthy controls was dominated by Firmicutes (69% and 76% of OTUs classifiable to the phylum level, respectively), followed by Bacteroidetes (10% and 6%), Actinobacteria (9% and 10%), Fusobacteria (6% and 3%), Proteobacteria (5% and 4%), and <1%, Spirochaetes, TM7, SR1, and Tenericutes (Fig. 2 B and C). The NSC analysis did not reveal any species-level OTUs that could discriminate between the healthy and patient oral samples.

To search for correlations between the abundances of OTUs in the oral cavity and markers for cardiovascular disease, we required a minimum sequence count of 100 sequences per genus (across all samples, using 1,700 sequences randomly selected per sample) for inclusion in the analysis. This analysis revealed several medically important OTUs in the oral cavity whose abundances correlated markers for cardiovascular disease (Fig. 4A). The abundance of Fusobacterium was positively correlated with levels of cholesterol and LDL cholesterol (ρ = 0.63 and 0.75, P = 0.028 and 0.005, respectively). Interestingly, the abundance of Streptococcus was positively correlated to HDL cholesterol and ApoAI levels (ρ = 0.88 and 0.70, P = 0.0001 and 0.01), whereas the abundance of Neisseria was negatively correlated to levels of these two disease markers (ρ = –0.65 and –0.74, P = 0.02 and 0.005).

Fig. 4.
Correlations between the abundances of different genera and disease markers in oral and fecal samples. Pearson correlation coefficients are represented by color ranging from blue, negative correlation (−1), to red, positive correlation (1). ( ...

Composition of the Gut Microbiota and Its Relation to Atherosclerosis and Disease Markers.

Because the fecal microbiota is representative of the gut microbiota (1), we characterized the bacterial diversity of fecal samples to determine whether gut microbiotas differed between patients and controls. Overall, the relative abundances of the phyla were similar between patients and controls, and NSC analysis did not reveal any OTUs whose abundances could differentiate the patients from the controls (Fig. 2). Compared with the oral and atherosclerotic plaque samples, gut samples contained significantly greater abundances of OTUs classified as members of the Lachnospiraceae family, and as the genera Ruminococcocus and Faecalibacterium. We found the abundances of two OTUs classified as uncharacterized members of Erysipelotrichaceae and Lachnospiraceae families in the gut to be positively correlated with cholesterol (ρ = 0.69 and ρ = 0.79, P = 0.009 and 0.001, respectively) and LDL cholesterol levels (ρ = 0.67 and ρ = 0.70, P = 0.012 and 0.007, respectively). In addition, we observed several other OTUs that correlated with disease markers but did not reach significance (Fig. 4B).

Inter- and Intraindividual Comparisons of Atherosclerotic Plaque, Oral, and Gut Microbiotas.

One of the main purposes of the study was to search for OTUs shared between oral and atherosclerotic plaque samples, and gut and atherosclerotic plaque samples, within the same individuals. Table 2 summarizes the OTUs that could be found in atherosclerotic plaques and at least one other body habitat in at least two patients. We detected Veillonella OTUs in all 13 patients (Table 2). In 11 of the 13 patients that provided oral samples, these OTUs could be detected in both the atherosclerotic plaques and the oral cavity samples of the same patients, and in two patients these OTUs were also detected in the gut. Streptococcus OTUs had a similar pattern of distribution: In six of the 10 patients, Streptococcus OTUs were detected in the oral cavity and atherosclerotic plaques, and in four patients they could also be detected in the gut. Interestingly, the combined abundances of Veillonella and Streptococcus were correlated in oral cavity and atherosclerotic plaque across patients (ρ = 0.6, P = 0.03), although when considered singly, the abundances of these bacteria in oral cavity and atherosclerotic plaque were not correlated.

Table 2.
OTUs shared among at least two body sites within the same patient

Within individual patients, the atherosclerotic plaque samples contained additional OTUs that were also detected in oral and gut samples. For instance, additional OTUs detected in the atherosclerotic plaque and oral samples of the same individual for at least two patients include Propionibacterium, Rothia, Burkholderia, Corynebacterium, Granulicatella, Staphylococcus and an unclassified OTU belonging to the Betaproteobacteria. In contrast, OTUs detected in the atherosclerotic plaque and gut samples of the same individual for at least two patients include Bacteroides, an unclassified member of the Lachnospiraceae, Bryantella, Enterobacter, an unclassified Enterobacteriaceae, Ruminococcus, and OTUs classified as Subdoligranulum (Table 2).


Here we used barcoded multiplexed-454 pyrosequencing to compare the bacterial composition of the oral, gut, and atherosclerotic plaque microbiotas in patients with atherosclerosis. This approach allowed a relatively comprehensive description of microbial communities associated with atherosclerotic plaques. This study also provides the interindividual comparison of oral, gut, and atherosclerotic plaque samples necessary to identify members of the normal microbiome that may translocate from one body habitat to another where they may contribute to disease. We found that the atherosclerotic plaque microbial communities are like those colonizing other parts of the human body—represented by a core of “signature” species, yet highly diverse and variable between individuals (4). Our findings suggest that the atherosclerotic plaque microbiota may at least in part be derived from the oral cavity and/or the gut. Although we did not observe any overall differences in the oral or gut microbiotas between patients and controls, several taxa correlated with cholesterol and apolipoprotein levels in the plasma.

Chryseomonas (recently reclassified as P. luteola) was present in all atherosclerotic plaque samples and may represent a previously unappreciated core member of atherosclerotic plaque communities. This organism has been implicated in endocarditis (24), and its presence in all atherosclerotic plaques studied here raises the possibility that it may directly affect the pathogenesis of atherosclerosis. Although we did not detect it from buccal swabs, it has been described as an inhabitant of the tongue (25). In addition to the diversity described here, another important constituent of atherosclerotic plaque is Chlamydia (26). Although not detected in the 16S rRNA gene diversity survey [likely due to suboptimal hybridization of the 16S rRNA gene primers we used to members of this phylum (Chlamydiae) (17)], we confirmed its presence in all plaques by qPCR with Chlamydia-specific primers (Table S2).

One intriguing observation was the correlation of combined Veillonella and Streptococcus abundances in the atherosclerotic plaque samples and in the oral samples across patients. These two genera are known early colonizers of tooth surfaces that interact to form dental plaques: Streptococcus produces a preferred fermentation product for Veillonella (27). However, because both genera also are common colonizers of the stomach and small intestine (28, 29), we cannot rule out that they may originate from these body sites. Besides these OTUs, we also detected Rothia, Granulicatella, and Propionibacterium in both the oral cavity and the atherosclerotic plaque within the same patients. All three of these genera have been described as normal inhabitants of the healthy oral microbiota (30). Furthermore, they have also been implicated in endocarditis (3133), and recent data suggest that the transmission of Granulicatella from the oral cavity could cause infective endocarditis (31). Interestingly, the oral and atherosclerotic plaque microbiotas have similar levels of diversity (PD). Thus, the pathobiology of the atherosclerotic plaque may be similar to that of dental plaque—a polymicrobial infection characterized by a few key members required for the initiation of disease.

Our analysis also revealed several OTUs shared between the atherosclerotic plaque and the gut, suggesting that bacteria present in the atherosclerotic plaque could also be derived from the distal gut as well as the oral cavity. One mechanism by which bacteria could reach the atherosclerotic plaque is phagocytosis by macrophages at epithelial linings (e.g., the oral cavity, gut, and the lung). Upon phagocytosis, the macrophages become activated, and when they reach the activated endothelium of the atheroma, they leave the blood stream to enter the atheroma and transform into cholesterol-laden foam cells (26). In support of this mechanism, patients with cardiovascular disease have a twofold increase of C. pneumonia-infected peripheral blood mononuclear cells compared with controls (34). Furthermore, bacteria are only present in atheromas and not in healthy aortic tissues in mice (35) and have been identified in human atherosclerotic plaques (17). Thus, infected macrophages may specifically target bacteria to atheromas.

An accumulating body of data indicates that the impact of infection on atherosclerosis is related to the total “pathogen burden,” i.e., the aggregate number of pathogens infecting an individual (13). Indeed, the observed correlation between the amount of bacterial DNA in atherosclerotic plaques and the leukocyte counts further suggests that atherosclerotic plaque bacterial load determines its inflammatory status and stability. Our findings may thus suggest that bacteria predominantly affect atherosclerosis by activating the innate immune system in the atherosclerotic plaque. Accordingly, atherosclerosis-prone mice deficient in Toll-like receptor (Tlr) 2, Tlr4, or the adapter molecule MyD88 are resistant to the development of atherosclerosis (3638). Furthermore, polymorphism in Tlr4, which promotes production of proinflammatory cytokines in response to bacterial LPS, is associated with lower levels of proinflammatory cytokines, acute-phase reactants, carotid atherosclerosis, and a smaller intima-media thickness in the common carotid artery (39).

In addition to the shared OTUs between atherosclerotic plaques and oral/gut microbiotas, our analysis revealed that specific components of the oral and gut microbiotas correlated with disease markers. Streptococcus was strongly positively correlated with HDL cholesterol and ApoAI (a major component of HDL), whereas Neisseria was strongly negatively correlated with these markers. Fusobacterium abundance was positively correlated with LDL cholesterol and total cholesterol. Similarly, members of the Erysipelotrichaceae and Lachnospiraceae families in the gut also positively correlated with LDL cholesterol and total cholesterol. Whereas the oral components are unlikely to directly contribute to cholesterol metabolism, they may serve as biomarkers. In contrast, the gut microbiota may directly affect host lipid metabolism (40). Although the exact mechanism remains to be identified, the gut microbiota is essential for bioconversion of cholesterol to bile acids, which are required for cholesterol excretion, as well as cholesterol absorption (4143). Accordingly, in addition to promoting atherosclerotic plaque inflammatory status, the gut microbiota could also contribute to atherosclerotic disease by modulating host lipid metabolism.

In summary, we detected key bacterial members of dental plaque in atherosclerotic plaques in humans, as well as a novel common member, Chryseomonas, in all atherosclerotic plaques. In addition, the atherosclerotic plaques contained numerous bacteria from different phyla. Our findings strongly support the hypothesis that the oral cavity and gut can be sources for atherosclerotic plaque-associated bacteria. Furthermore, members of both the oral and gut microbiotas correlated with disease biomarkers, especially plasma cholesterol. Based on these findings, several lines of inquiry are warranted: investigation into whether oral and/or fecal bacterial community composition can predict atherosclerosis, and how specific bacteria, such as Chryseomonas, may contribute to atherosclerosis development and/or progression.

Materials and Methods

Patient and Control Groups.

The patient samples were obtained from the Göteborg Atheroma Study Group biobank, which includes carotid endarterectomies from patients who were operated for minor ischemic stroke, transient ischemic attack, or amaurosis fugax as previously described in detail (44). The patients were consecutively included, and completed questionnaires covering previous and current diseases, lifestyle factors, and medication. Before surgery, blood samples were drawn and plasma and serum aliquots were prepared and immediately frozen at −70 °C. The excised endarterectomy specimens were immediately frozen in liquid nitrogen under sterile conditions. The subjects in the control group were obtained from two currently running population-based studies of men and women born 1937–1940 (4547). These studies were based on screening examinations of randomly selected population-based cohorts. The control subjects were matched to the patient group by sex, and had to fulfill criteria of feeling well and not suffering from past or current cerebrovascular disease. Each control subject came to the laboratory for information and examinations identical to those performed in the patient group. Mouth swabs were obtained from both groups by a nurse, and the patients were given material and instructions for providing fecal samples. Of the complete sample set, two samples are missing: one oral sample (not obtained) and one atherosclerotic plaque (insufficient material). The study was approved by the ethics committee in Gothenburg. All patients gave written informed consent to participate after oral and written information.

DNA Extraction.

Genomic DNA was isolated from 100 mg of feces using the Viogene DNA Extraction Kit. Mouth swabs were soaked in 900 μL lysis buffer for 2 h before DNA was isolated using the same kit. The genomic DNA from the atherosclerotic plaque samples was extracted using the MOBIO PowerSoil DNA Isolation Kit. All samples were extensively homogenized using a bead beater at maximum speed for 3 min. The remaining steps were performed as directed by the manufacturer.


The bacterial DNA concentration in plaque was quantified by quantitative PCR (qPCR) according to Sokol et al. (48) and performed under the following conditions: 50 °C for 2 min, 95 °C for 10 min, 45 cycles of 95 °C for 15 s, and 62 °C for 1 min. A 1:1 mix of purified genomic DNA from Escherichia coli and Lactobacillus reuteri was used to construct a standard curve in the range of 10 fg to 10 ng. The amount of Chlamydia spp. was quantified according to Yoshida et al. (49) and performed under the following conditions: 95 °C for 10 min, 40 cycles of 95 °C for 15 s, and 56 °C for 1 min. A standard curve was made from genomic DNA from Chlamydia trachomatis (DNA from cells infected with C. trachomatis, serovar H, strain UW-43/Cx, ATCC, VR-879D) in the range of 10 fg to 10 ng. To ensure that the primers were specific for Chlamydia, negative control reactions with E. coli and Bifidobacterium longum DNA were performed.

PCR Amplification of the V1-2 Region of Bacterial 16S rRNA Genes.

16S rRNA genes were amplified by PCR with a forward primer containing the 454 Life Sciences primer B sequence and the broadly conserved bacterial primer 27F, and a reverse primer containing the 454 Life Sciences primer A sequence, a unique 12-nucleotide error-correcting barcode used to tag each PCR product, and the broad-range bacterial primer 338R (22). PCR reactions were carried out in quadruplicate (gut, oral) or triplicate (atherosclerotic plaques) 20-μL reactions with 0.3 μM forward and reverse primers, with ≈50 ng template DNA and 1× of HotStar Taq Plus Master Mix kit (Qiagen). Thermal cycling consisted of initial denaturation at 95 °C for 2 min followed by 30 cycles of denaturation at 95 °C for 20 s, annealing at 52 °C for 20 s, and extension at 65 °C for 60 s. Replicate amplicons were pooled, purified with Agencourt AMPure Kit (Agencourt), and visualized by electrophoresis using 1.0% agarose gels. Negative extraction control samples were treated similarly, except that no template DNA was added to the PCR reactions; these failed to produce visible PCR products and were not analyzed further.

Amplicon Quantitation, Pooling, and Pyrosequencing.

Oral and fecal amplicon DNA concentrations were determined by spectrophotometry (Nanodrop), and atherosclerotic plaque amplicons DNAs were quantified using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen). Amplicons were combined in equimolar ratios into a single tube with a final concentration of 16 ng/μL. Pyrosequencing of oral and fecal samples was carried on a 454 Life Sciences Genome Sequencer FLX instrument at Center for Metagenomic Sequence Analysis at KTH, School of Biotechnology in Stockholm, Sweden, and atherosclerotic plaque samples were sequenced at the Cornell University Life Sciences Core Laboratories Center using Roche/454 Titanium chemistry.

Sequence Analysis.

Sequences were denoised and analyzed with the software package Quantitative Insights into Microbial Ecology (QIIME) using default parameters for each step, except where specified (50). Sequences were removed if lengths were <200 nt; contained ambiguous bases, primer mismatches, homopolymer runs in excess of six bases, or uncorrectable barcodes; or lacked the primer. Remaining sequences were assigned to samples according to their barcodes (Table S1). Similar sequences were binned into OTUs using UCLUST (http://www.drive5.com/usearch/), with a minimum pairwise identity of 97%. The most abundant sequence in each OTU was chosen to represent its OTU. Representative sequences from each OTU were aligned using PyNAST (a python-based implementation of NAST in QIIME) (51) and the Greengenes (52) database (corset aligned November 8, 2007) using a minimum percent identity of 75%. Representative sequences were aligned and ChimeraSlayer (http://microbiomeutil.sourceforge.net/) was used for the identification of putative chimeric sequences. From the putatively chimeric sequences in the representative set, we counted the number of total sequences in each of the corresponding OTUs to determine the proportion of total sequences deemed putative chimeras. The lanemaskPH was used to screen out the hypervariable regions (4), and a phylogenetic tree was constructed using FastTree (53). Taxonomy was assigned using the Ribosomal Database Project (RDP) classifier (54) with a minimum support threshold of 60% and the RDP classifier nomenclature. Sequences belonging to OTUs classified as plant chloroplasts were considered contaminants and removed from the analysis.

To compare diversity between samples, we used the weighted and unweighted UniFrac distance metrics (55, 56) using a random sample of 1,700 sequences per sample (one atherosclerotic plaque sample had fewer than 1,500 sequences and was omitted from the analysis). To relate the OTU abundances to patient health data, we considered only OTUs containing at least 100 sequences.

Statistical Analyses.

A “nearest-shrunken centroid” classification approach was performed to detect the genera that were particularly representative of each category (atherosclerotic plaque, oral cavity, and gut for the different body habitats, and healthy vs. patient for within oral cavity and gut) (57). The amount of shrinkage was set to minimize the misclassification error. When phyla were used to classify samples, the accuracy of the classification was quite good (overall error rate = 0.097; only 7 of 72 samples were misclassified). All of the misclassified samples were oral cavity samples incorrectly classified as gut. At the genus level, the classification improved (overall error rate = 0.014; one atherosclerotic plaque sample was misclassified as gut). These analyses allowed the identification of phyla/genera whose abundances significantly differed between categories. This analysis was performed using the Prediction Analysis for Microarrays (PAM) package within R software.

Supplementary Material

Supporting Information:


We are thankful to our colleagues of the Göteborg Atheroma Study Group (Sahlgrenska University Hospital and University of Gothenburg) who provided endarterectomies, serum samples, and clinical data; Kicki Holmberg at the Center for Metagenomic Sequence Analysis for sequencing; and Marie-Louise Ekholm, Paerhati Aibibula, and Nick Scalfone for skillful technical assistance. This work was supported by the Swedish Research Council, Swedish Foundation for Strategic Research, Torsten and Ragnar Söderberg's Foundation, Petrus and Augusta Hedlund's Foundation, the Swedish federal government under the Swedish Heart Lung Foundation and the health authorities of the Västra Götaland Region (LUA/ALF) agreement, the Howard Hughes Medical Institute, National Institutes of Health Grants HG4872 and HG4866, the National Institutes of Health Human Microbiome Project Data Analysis and Coordination Center (O.K.), National Science Foundation Grant 0958184, the Beckman Young Investigator Program, and The Hartwell Foundation (R.E.L.).


The authors declare no conflict of interest.

This paper results from the Arthur M. Sackler Colloquium of the National Academy of Sciences, “Microbes and Health,” held November 2–3, 2009, at the Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering in Irvine, CA. The complete program and audio files of most presentations are available on the NAS Web site at http://www.nasonline.org/SACKLER_Microbes_and_Health.

This article is a PNAS Direct Submission.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1011383107/-/DCSupplemental.


1. Eckburg PB, et al. Diversity of the human intestinal microbial flora. Science. 2005;308:1635–1638. [PMC free article] [PubMed]
2. Dethlefsen L, Eckburg PB, Bik EM, Relman DA. Assembly of the human intestinal microbiota. Trends Ecol Evol. 2006;21:517–523. [PubMed]
3. Turnbaugh PJ, et al. A core gut microbiome in obese and lean twins. Nature. 2009;457:480–484. [PMC free article] [PubMed]
4. Costello EK, et al. Bacterial community variation in human body habitats across space and time. Science. 2009;326:1694–1697. [PMC free article] [PubMed]
5. Grice EA, et al. NISC Comparative Sequencing Program Topographical and temporal diversity of the human skin microbiome. Science. 2009;324:1190–1192. [PMC free article] [PubMed]
6. Nasidze I, Li J, Quinque D, Tang K, Stoneking M. Global diversity in the human salivary microbiome. Genome Res. 2009;19:636–643. [PMC free article] [PubMed]
7. Savage DC. Microbial ecology of the gastrointestinal tract. Annu Rev Microbiol. 1977;31:107–133. [PubMed]
8. Duncan SH, et al. Human colonic microbiota associated with diet, obesity and weight loss. Int J Obes (Lond) 2008;32:1720–1724. [PubMed]
9. Schwiertz A, et al. Microbiota and SCFA in lean and overweight healthy subjects. Obesity (Silver Spring) 2010;18:190–195. [PubMed]
10. Bäckhed F, et al. The gut microbiota as an environmental factor that regulates fat storage. Proc Natl Acad Sci USA. 2004;101:15718–15723. [PMC free article] [PubMed]
11. Bäckhed F, Manchester JK, Semenkovich CF, Gordon JI. Mechanisms underlying the resistance to diet-induced obesity in germ-free mice. Proc Natl Acad Sci USA. 2007;104:979–984. [PMC free article] [PubMed]
12. Hansson GK. Inflammation, atherosclerosis, and coronary artery disease. N Engl J Med. 2005;352:1685–1695. [PubMed]
13. Epstein SE, et al. Infection and atherosclerosis: Potential roles of pathogen burden and molecular mimicry. Arterioscler Thromb Vasc Biol. 2000;20:1417–1420. [PubMed]
14. Mattila KJ, et al. Association between dental health and acute myocardial infarction. BMJ. 1989;298:779–781. [PMC free article] [PubMed]
15. Saikku P, et al. Chronic Chlamydia pneumoniae infection as a risk factor for coronary heart disease in the Helsinki Heart Study. Ann Intern Med. 1992;116:273–278. [PubMed]
16. Holmes E, et al. Human metabolic phenotype diversity and its association with diet and blood pressure. Nature. 2008;453:396–400. [PubMed]
17. Ott SJ, et al. Detection of diverse bacterial signatures in atherosclerotic lesions of patients with coronary heart disease. Circulation. 2006;113:929–937. [PubMed]
18. Haraszthy VI, Zambon JJ, Trevisan M, Zeid M, Genco RJ. Identification of periodontal pathogens in atheromatous plaques. J Periodontol. 2000;71:1554–1560. [PubMed]
19. Kozarov EV, Dorn BR, Shelburne CE, Dunn WA, Jr, Progulske-Fox A. Human atherosclerotic plaque contains viable invasive Actinobacillus actinomycetemcomitans and Porphyromonas gingivalis. Arterioscler Thromb Vasc Biol. 2005;25:e17–e18. [PubMed]
20. Stelzel M, et al. Detection of Porphyromonas gingivalis DNA in aortic tissue by PCR. J Periodontol. 2002;73:868–870. [PubMed]
21. Gaetti-Jardim E, Jr, Marcelino SL, Feitosa AC, Romito GA, Avila-Campos MJ. Quantitative detection of periodontopathic bacteria in atherosclerotic plaques from coronary arteries. J Med Microbiol. 2009;58:1568–1575. [PubMed]
22. Hamady M, Walker JJ, Harris JK, Gold NJ, Knight R. Error-correcting barcoded primers for pyrosequencing hundreds of samples in multiplex. Nat Methods. 2008;5:235–237. [PMC free article] [PubMed]
23. Faith DP. Conservation evaluation and phylogenetic diversity. Biol Conserv. 1992;61:1–10.
24. Casalta JP, Fournier PE, Habib G, Riberi A, Raoult D. Prosthetic valve endocarditis caused by Pseudomonas luteola. BMC Infect Dis. 2005;5:82. [PMC free article] [PubMed]
25. Conti S, dos Santos SS, Koga-Ito CY, Jorge AO. Enterobacteriaceae and pseudomonadaceae on the dorsum of the human tongue. J Appl Oral Sci. 2009;17:375–380. [PMC free article] [PubMed]
26. Campbell LA, Kuo CC. Chlamydia pneumoniae—an infectious risk factor for atherosclerosis? Nat Rev Microbiol. 2004;2:23–32. [PubMed]
27. Chalmers NI, Palmer RJ, Cisar JO, Kolenbrander PE. Characterization of a Streptococcus sp.-Veillonella sp. community micromanipulated from dental plaque. J Bacteriol. 2008;190:8145–8154. [PMC free article] [PubMed]
28. Andersson AF, et al. Comparative analysis of human gut microbiota by barcoded pyrosequencing. PLoS ONE. 2008;3:e2836. [PMC free article] [PubMed]
29. Macy JM, Yu I, Caldwell C, Hungate RE. Reliable sampling method for analysis of the ecology of the human alimentary tract. Appl Environ Microbiol. 1978;35:113–120. [PMC free article] [PubMed]
30. Bik EM, et al. Bacterial diversity in the oral cavity of 10 healthy individuals. ISME J. 2010;4:962–974. [PMC free article] [PubMed]
31. Ohara-Nemoto Y, et al. Infective endocarditis caused by Granulicatella elegans originating in the oral cavity. J Clin Microbiol. 2005;43:1405–1407. [PMC free article] [PubMed]
32. Ricaurte JC, et al. Rothia dentocariosa endocarditis complicated by multiple intracranial hemorrhages. South Med J. 2001;94:438–440. [PubMed]
33. Clayton JJ, Baig W, Reynolds GW, Sandoe JA. Endocarditis caused by Propionibacterium species: A report of three cases and a review of clinical features and diagnostic difficulties. J Med Microbiol. 2006;55:981–987. [PubMed]
34. Smieja M, Mahony J, Petrich A, Boman J, Chernesky M. Association of circulating Chlamydia pneumoniae DNA with cardiovascular disease: A systematic review. BMC Infect Dis. 2002;2:21. [PMC free article] [PubMed]
35. Moazed TC, Kuo CC, Grayston JT, Campbell LA. Murine models of Chlamydia pneumoniae infection and atherosclerosis. J Infect Dis. 1997;175:883–890. [PubMed]
36. Mullick AE, Tobias PS, Curtiss LK. Modulation of atherosclerosis in mice by Toll-like receptor 2. J Clin Invest. 2005;115:3149–3156. [PMC free article] [PubMed]
37. Björkbacka H, et al. Reduced atherosclerosis in MyD88-null mice links elevated serum cholesterol levels to activation of innate immunity signaling pathways. Nat Med. 2004;10:416–421. [PubMed]
38. Michelsen KS, et al. Lack of Toll-like receptor 4 or myeloid differentiation factor 88 reduces atherosclerosis and alters plaque phenotype in mice deficient in apolipoprotein E. Proc Natl Acad Sci USA. 2004;101:10679–10684. [PMC free article] [PubMed]
39. Kiechl S, et al. Toll-like receptor 4 polymorphisms and atherogenesis. N Engl J Med. 2002;347:185–192. [PubMed]
40. Velagapudi VR, et al. The gut microbiota modulates host energy and lipid metabolism in mice. J Lipid Res. 2010;51:1101–1112. [PMC free article] [PubMed]
41. Gustafsson BE, Midtvedt T, Norman A. Isolated fecal microorganisms capable of 7-alpha-dehydroxylating bile acids. J Exp Med. 1966;123:413–432. [PMC free article] [PubMed]
42. Gustafsson BE, Norman A. Comparison of bile acids in intestinal contents of germfree and conventional rats. Proc Soc Exp Biol Med. 1962;110:387–389. [PubMed]
43. Wostmann BS. Intestinal bile acids and cholesterol absorption in the germfree rat. J Nutr. 1973;103:982–990. [PubMed]
44. Fagerberg B, et al. Differences in lesion severity and cellular composition between in vivo assessed upstream and downstream sides of human symptomatic carotid atherosclerotic plaques. J Vasc Res. 2010;47:221–230. [PubMed]
45. Bokemark L, Wikstrand J, Wedel H, Fagerberg B, Atherosclerosis and Insulin Resistance sstudy (AIR) Insulin, insulin propeptides and intima-media thickness in the carotid artery in 58-year-old clinically healthy men. The Atherosclerosis and Insulin Resistance study (AIR) Diabet Med. 2002;19:144–151. [PubMed]
46. Brohall G, Behre CJ, Hulthe J, Wikstrand J, Fagerberg B. Prevalence of diabetes and impaired glucose tolerance in 64-year-old Swedish women: Experiences of using repeated oral glucose tolerance tests. Diabetes Care. 2006;29:363–367. [PubMed]
47. Sigurdardottir V, Fagerberg B, Hulthe J. Preclinical atherosclerosis and inflammation in 61-year-old men with newly diagnosed diabetes and established diabetes. Diabetes Care. 2004;27:880–884. [PubMed]
48. Sokol H, et al. Faecalibacterium prausnitzii is an anti-inflammatory commensal bacterium identified by gut microbiota analysis of Crohn disease patients. Proc Natl Acad Sci USA. 2008;105:16731–16736. [PMC free article] [PubMed]
49. Yoshida H, Kishi Y, Shiga S, Hagiwara T. Differentiation of Chlamydia species by combined use of polymerase chain reaction and restriction endonuclease analysis. Microbiol Immunol. 1998;42:411–414. [PubMed]
50. Caporaso JG, et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods. 2010;7:335–336. [PMC free article] [PubMed]
51. Caporaso JG, et al. PyNAST: A flexible tool for aligning sequences to a template alignment. Bioinformatics. 2010;26:266–267. [PMC free article] [PubMed]
52. DeSantis TZ, et al. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl Environ Microbiol. 2006;72:5069–5072. [PMC free article] [PubMed]
53. Price MN, Dehal PS, Arkin AP. FastTree: Computing large minimum evolution trees with profiles instead of a distance matrix. Mol Biol Evol. 2009;26:1641–1650. [PMC free article] [PubMed]
54. Cole JR, et al. The ribosomal database project (RDP-II): Introducing myRDP space and quality controlled public data. Nucleic Acids Res. 2007;35(Database issue):D169–D172. [PMC free article] [PubMed]
55. Lozupone C, Knight R. UniFrac: A new phylogenetic method for comparing microbial communities. Appl Environ Microbiol. 2005;71:8228–8235. [PMC free article] [PubMed]
56. Lozupone C, Hamady M, Knight R. UniFrac—an online tool for comparing microbial community diversity in a phylogenetic context. BMC Bioinformatics. 2006;7:371. [PMC free article] [PubMed]
57. Tibshirani R, Hastie T, Narasimhan B, Chu G. Diagnosis of multiple cancer types by shrunken centroids of gene expression. Proc Natl Acad Sci USA. 2002;99:6567–6572. [PMC free article] [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...