Metagenomic analysis of oral plaques and aortic valve tissues reveals oral bacteria associated with aortic stenosis

Bacteria derived from the oral cavity enter the bloodstream and cause the onset of various systemic diseases, including heart valve disease. However, information on the oral bacteria involved in aortic stenosis is limited. We comprehensively analyzed the microbiota in aortic valve tissues collected from aortic stenosis patients using metagenomic sequencing and investigated the relationships between the valve microbiota, the oral microbiota, and oral cavity conditions. Metagenomic analysis revealed the presence of 629 bacterial species in five oral plaques and 15 aortic valve clinical specimens. Patients were classified into two groups (A and B) according to their aortic valve microbiota composition using principal coordinate analysis. Examination of the oral conditions of the patients showed no difference in the decayed/missing/filled teeth index. Bacteria in group B tend to be associated with severe disease, and the number of bacteria on the dorsum of the tongue and the positive rate of bleeding during probing were significantly higher in this group than in group A. The pathophysiology of aortic stenosis may be related to the presence of oral bacteria such as Streptococcus oralis and Streptococcus sanguinis following bacteremia. Systemic inflammation in severe periodontitis may be driven by the oral microbiota, supporting the indirect (inflammatory) association between oral bacteria and aortic stenosis. Appropriate oral hygiene management may contribute to the prevention and treatment of aortic stenosis.


Introduction
Numerous indigenous microorganisms exist in the human body, forming a microbiota. Microbiota exists on the entire body skin, respiratory mucosae (including the oral cavity, nasal cavity, and alveoli), and digestive system components (including the esophagus, small intestine, and large intestine) and the genitourinary system (including the uterus, vagina, bladder, and ureter). With approximately 700 different species, the oral microbiota is one of the most complex microbial communities in the human body [1,2]. The oral microbiota protects the host from the intrusion of harmful pathogens from the external environment. However, microbiota dysbiosis has been reported to be closely related to dental and periodontal diseases, such as caries [3], periodontitis [4], oral mucosal diseases [5], oral cancer [6], and various systemic diseases [7], including gastrointestinal [8], neurological [9], and cardiovascular diseases [10]. Sayaka Yoshiba and Hirofumi Nakagawa contributed equally to this work. Accordingly, the oral microbiota plays an important role in the homeostasis of human health [11].
Oral bacteria are known to enter blood vessels during invasive procedures such as tooth extraction [12]. Bacteremia may occur during periodontal treatment, daily toothbrushing, and mastication, as oral bacteria may enter the circulation at relatively high frequencies [13,14], thus reaching various tissues and organs via the bloodstream where they may exert unexplained biological effects [13].
In 1891, Miller [15] suggested that oral infections are not confined to the oral cavity but spread throughout the body and proposed a focal infection theory focusing on periodontal pathogens as the cause of systemic infections. A study in 1989 showed that poor oral hygiene is associated with myocardial infarction [16]. Since then, it has been verified that periodontal disease is a risk factor for cardiovascular disease [17]. It has been reported that patients with coronary artery disease have deep periodontal pockets and increased antibody titers against Porphyromonas gingivalis and Prevotella intermedia [18], and that patients with peripheral vascular disease have overwhelmingly high rates of tooth loss [19]. Oral care and oral management, including periodontal disease treatment, may thus be effective in reducing cardiovascular disease, but this has not been validated yet.
Conventional culture-independent methods, such as DNA-DNA hybridization, DNA cloning, and sequencing, have been widely used to identify oral microbes [20]. However, these methods are biased due to incomplete coverage of the microbial diversity because they fail to detect numerous low-abundance bacterial species. Recently, metagenomic analysis has attracted attention as it can be used to comprehensively analyze extracted genomic DNA and reveal bacterial species composition and microbiota function [21]. This method enables the analysis of the entire genome of microbial communities without relying on culture methods, facilitating the identification of previously unknown species [22,23].
In aortic stenosis, the opening of the aortic valve narrows, impeding blood flow from the left ventricle to the aorta. As it progresses, various symptoms develop, including angina pectoris, syncope, and heart failure. Various etiologies, including congenital, inflammatory, ischemic, agingrelated, and degenerative processes, have been suggested to be related to the development of aortic stenosis [24,25]. However, the pathophysiology and progression of aortic stenosis remain largely unknown.
In this study, we focused on the involvement of dysbiosis in the oral microbiota in the homeostasis of oral and systemic physiological functions and the development of diseases. We performed metagenomic shotgun sequencing using the next-generation sequencer on DNA isolated from surgically resected valve tissues and oral plaques from aortic stenosis patients to study the relationship between the conditions of the oral cavity, the pathology, and the progression of aortic stenosis.

Study population
The study population comprised of 15 patients who received a diagnosis of aortic stenosis and underwent valve replacement at the Division of Cardiovascular Surgery, Showa University Northern Yokohama Hospital, Kanagawa, Japan, between November 2017 and August 2018. Written informed consent was obtained from all the enrolled individuals. To evaluate the oral cavity conditions, the number of remaining teeth, healthy teeth, treated teeth, and missing teeth, the decayed/missing/filled teeth (DMFT) index was determined. The pocket depth of the remaining teeth was measured using the one-point method, and the presence or absence of bleeding during probing was recorded. A periodontal probe (15 UNC Color-Coded Probe; Hu-Friedy, Chicago, IL, USA) was used for probing with a light pressure of 0.2-0.25 N. To measure the number of bacteria on the tongue dorsum, a sterile cotton swab was placed in a constant-pressure sample collection device, and the sample was collected by moving it forward from the center of the dorsum of the tongue in a width of 10 mm. Bacteria on the dorsum of the tongue were counted using a bacterial counter (Panasonic Healthcare Holdings Co., Ltd., Tokyo, Japan), which measures the change in impedance and converts these changes to a bacterial concentration to quantitatively measure bacteria colonies. Supragingival oral plaque samples were collected the day before surgery, and aortic valve tissues were collected during aortic valve replacement surgery. Since the number of remaining teeth varied among subjects, it was not possible to collect samples from the same tooth in all individuals. Therefore, samples were taken from the supragingival plaque of the remaining teeth. Clinical specimens were frozen until use. This study was approved by the Ethics Committee of Showa University (approval number 17H-049).

Metagenomic library preparation and sequencing
The amount of DNA collected from oral plaque differed from patient to patient; the quality of the collected DNA was evaluated and quantified. Since the size of the heart valve specimens collected differed based on the case, the size was unified to 10 mm × 10 mm pieces. Furthermore, the aortic valve pieces were cut into small pieces and homogenized using bead disruption prior to DNA extraction. Microbial genomic DNA was extracted using the QIAamp UCP Pathogen Mini Kit (QIAGEN, Hilden, Germany) as per the manufacturer's instructions. DNA quality and quantity were determined using 2200 TapeStation (Agilent, Santa Clara, CA, USA) and Invitrogen's Quant-iT™ dsDNA Broad-Range Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA), respectively. Libraries of all samples were prepared using the Rubicon ThruPLEX DNA-seq Kit (Rubicon Genomics, Ann Arbor, MI, USA) with a unique dual index adapter pair for each sample. The DNA libraries were sequenced in two lanes of an Illumina NovaSeq 6000 system (Illumina, Inc., San Diego, CA, USA) in the rapid run mode at a read length of 150 bp (paired-end), generating an average of 20.91 M (11.28-25.34 M) reads per sample. Metagenomic sequencing was performed at Takara Bio Inc. (Shiga, Japan).

Bioinformatic analysis of the microbiome
Trimmomatic was used to trim all paired-end reads in raw data (FASTQ files) to remove low-quality regions and short reads [26]. Host reads were removed by mapping all sequencing reads to the human reference genome using Bow-tie2 (v2.3.1) [27]. The reads were then aligned to NCBInr (ftp:// ftp. ncbi. nih. gov/ blast/ db/ FASA/ nr. gz database, downloaded on August 2019) using Kaiju [28]. Hits with NCBI Taxonomy ID were converted to taxon names using Tax-onKit [29], and the relative abundance of the microbial communities in each sample was calculated based on the number of mapped reads. Based on a taxonomic composition table, relative abundance was visualized in a taxon bar chart using QIIME2 [30]. Based on the microbiota composition, heart valve specimens were classified into two groups (A and B). The degree of similarity (distance) between the groups and oral plaque specimens was calculated using the Bray-Curtis dissimilarity and principal coordinate analysis (PCoA) with QIIME1 [31].

Polymerase chain reaction (PCR) detection of Streptococcus gordonii
Reaction mixtures (20 μL) containing 0.5 U TaKaRaEx Taq (TAKARA Bio, Inc., Otsu, Shiga, Japan), 0.5 μM oligonucleotide primers, 5 ng template DNA, and 1.5 mM MgCl 2 were prepared according to the manufacturer's protocol. Amplification was performed using the GeneAmp 9700 thermal cycler (Applied Biosystems, Foster City, CA, USA) under the following thermal cycling conditions: 30 cycles of denaturation at 98 °C for 10 s and 1 min each for primer annealing and extension at 70 °C. The PCR primers used have been previously reported [32]. The primer sequences used were 5′-CTA TGC GGA TGA TGC TAA TCA AGT G-3′ (sense) and 5′-GGA GTC GCT ATA ATC TTG TCA GAA A-3′ (antisense), and the expected product size was 440 bp. Streptococcus gordonii ATCC10558 was the positive control.

Statistical analysis
The relationship between clinical information and oral conditions was compared between groups A and B participants using the chi-squared test, t test, and ANOVA. Results with p < 0.05 were considered statistically significant.

Oral conditions of the study participants
The average numbers of remaining, healthy, treated, and missing teeth were 18.7, 8.5, 10.2, and 9.9, respectively. The average DMFT index was 0.72. Of the remaining teeth in the participants, 10.2% had pockets of 4 mm or more. The positive bleeding rate on probing of the remaining teeth was 6.8%. The average number of bacteria on the dorsum of the tongue was 1.88 × 10 7 .
The 10 most abundant bacterial genera identified are displayed in Fig. 1, upper panel. The most predominant genera detected in the oral plaque samples were Prevotella, Streptococcus, and Actinomyces, whereas the most abundant genera in the aortic valve specimens were Clostridioides, Corallococcus, and Nocardia. The 10 most predominant bacterial species are presented in Fig. 1, lower panel. The most predominant species detected in the oral plaque specimens were Actinomyces israelii, Corynebacterium matruchotii, and Olsenella sp. oral taxon 807, whereas the most abundant species in the aortic valve specimens were Clostridioides difficile, Corallococcus sp. AB049A, and Nocardia cyriacigeorgica.

Bacterial species commonly identified in oral plaques and heart valves
We identified 293 bacterial species in oral plaques and aortic valve tissues collected from five patients (Fig. 2). Among these, 197 species (67.2%) were specifically associated with oral plaque, 87 species (29.7%) were identified only in aortic valve tissue, and 9 species (3.1%) were shared between the oral plaques and aortic valve tissues. These nine shared species were Streptococcus oralis and Streptococcus sanguinis (common oral bacteria), Streptococcus agalactiae (β-hemolytic streptococcus), Streptococcus pneumoniae, Pseudomonas aeruginosa, Escherichia coli, Lactobacillus rhamnosus, Ralstonia solanacearum, and C. difficile.

Verification of bacterial species identified using metagenomic analysis
Based on the results of the metagenomic analysis, the oral bacteria S. oralis (average detection rate, 8.52E − 04%) and S. gordonii (average detection rate, 83.15E − 04%) were found in all 15 heart valve specimens. The metagenomic analysis results for S. gordonii were verified using PCR with bacterial DNA extracted from the oral plaque and aortic valve tissues. The PCR results verified the presence of S. gordonii in both specimen types (Fig. 3).

Bacterial community composition by sample type
We used PCoA to determine whether the microbiota composition is influenced by the participant or the sample type (Fig. 4). The microbiota identified in the oral plaque  samples was clearly distinct from that identified in the aortic valve specimens. Based on the microbiota composition, the heart valve specimens could be classified into two groups, groups A and B. In group A, Corallococcus sp. AB049A, Micrococcus luteus, and Staphylococcus equorum were the most abundant bacterial species, whereas in group B, Nocardia cyriacigeorgica was the most abundant bacterial species (Table 1).

Comparison of clinical information between groups A and B
Twelve out of 15 cases (80%) had severe stenosis, and the other 3 cases (20%) had moderate stenosis. Thus, the aortic stenosis results are highly comparable (Table 2). We compared groups A and B in terms of age, sex, aortic valve disease stage, aortic orifice area, comorbidities, and serum albumin and brain natriuretic peptide (BNP) levels ( Table 2). In comparison to group A bacteria, bacteria belonging to group B were significantly correlated with low levels of serum albumin (p = 0.0026). Bacteria in group B tend to be associated, albeit not significantly, with severe disease classification and an aortic valve area of less than 1 cm 2 .
We then compared the oral cavity conditions, including the DMFT index, the number of bacteria on the dorsum of the tongue, and bleeding during probing, among the groups. Patients in group B had significantly higher bacterial counts on the dorsum of the tongue and a higher rate of bleeding on probing (Fig. 5).

Discussion
The communities of microorganisms that exist symbiotically within our bodies are called the microbiota. The number of bacterial cells in our body is the same or may exceed the   [32,33]. In this study, using next-generation sequencing and metagenomic analysis, we identified 13 phyla, 92 genera, and 410 bacterial species in five oral plaque specimens. In line with previous study findings, Bacteroidetes (29.1%), Firmicutes (25.4%), Actinobacteria (16.6%), Proteobacteria (13.2%), Fusobacteria (4.3%), and Spirochetes (1.2%), which accounted for 89.9% of the total bacterial phyla, were detected in our study. The oral cavity contains a very large variety of bacterial species as it is exposed to the external environment and due to the presence of various niche habitats, including hard and soft tissues, in the complex intra-oral environment. Segata et al. compared the bacterial composition of specimens from different parts of the oral cavity of healthy individuals and identified the following parts where oral bacterial communities formed: (1) tooth surface (supragingival plaque and subgingival plaque), (2) dorsum of the tongue (classified into the pharynx, palatine tonsils, and saliva), (3) buccal mucosa, keratinized gingiva, and hard palate [34]. According to their report, Actinobacteria, Gammaproteobacteria, Betaproteobacteria, and Flavobacteria are hallmark classes of supragingival plaque. Consistent with their findings, we detected Actinobacteria (average detection rate, 1.43E − 01%), Gammaproteobacteria (2.40E − 02%), Betaproteobacteria (8.32E − 02%), and Flavobacteriia (5.00E − 02%).
Dental and other procedures that commonly cause bleeding, such as tooth extraction, periodontal surgery, scaling, and root planning, additionally are assumed to contribute in the development of bacteremia [12]. Bacteremia has also been reported to occur after routine oral hygiene procedures such as brushing (0-26%), flossing (20-58%), and even chewing (17-51%) [13,14]. We identified 10 bacterial phyla, 69 bacterial genera, and 219 bacterial species in 15 aortic valve specimens. In particular, nine bacterial species (3.1%) were found to be in common between the oral plaque and aortic valve tissue samples (Fig. 2). Among these nine species, S. oralis, S. sanguinis, and S. pneumoniae belong to the mitis group of the genus Streptococcus. In previous studies involving 239 cases of infective endocarditis, mitis group species were identified in 61 (25.5%) of 100 (41.8%) cases in which oral streptococcus was the causative organism [35][36][37]. Our results also demonstrated that various bacterial species, including oral Streptococci, can frequently enter the bloodstream and colonize and invade cardiovascular tissues, exerting various effects.
Aortic stenosis is a pathological condition marked by the infiltration of inflammatory macrophages and the formation of extracellular matrix in the valve tissue. As these pathological conditions progress, the valve cusps thicken and harden, and mineral deposition occurs [38]. Anatomically, the aortic valve comprises of three layers: the collagen-rich fibrosa layer on the aortic side of the valve, the glycosaminoglycanrich spongiosa layer in the middle, and the elastin-rich ventricularis layer on the left ventricular side [39]. The onset site of aortic valve calcification in this three-layer structure is thought to be proximal to the base of the fibrosa layer, which is in contact with the aortic wall, and the calcification progresses from the base to the tip [40]. Additionally, pathological conditions or external stimulations promote the production of cytoskeletal or osteogenic proteins, which in turn induces the differentiation of valvular interstitial cells, which make up the aortic valve, into myofibroblasts and osteoblasts. It has been suggested that myofibroblasts promote fibrosis and sclerosis of the valve tissue, whereas osteoblasts promote the deposition of mineral components such as calcium [41]. However, the activation mechanism for cells with these varying properties in vivo has not been elucidated, and it is unknown whether the activation of these cells contributes to valve dysfunction. Oral streptococcal virulence factors include surface protein antigen I/II, serotype-specific polysaccharides, and lipoteichoic acid, all of which have been shown to induce inflammatory responses in human cells [42]. In addition, Streptococcus oralis expresses coaggregation receptor polysaccharide (RPS) that functions as a receptor for surface adhesins during symbiosis with other bacterial species, facilitating the formation of an oral biofilm. Recently, RPS has been reported to induce an inflammatory response in human aortic endothelial cells, leading to an increased production of cytokines, adhesion molecules, and TLR-2 [43]. Thus, these molecular mechanisms may be involved in the phenotype associated with valvular interstitial cells and the pathogenesis of aortic stenosis. Based on the characteristics of the microbiota, the aortic valve tissue samples were divided into two groups, groups A and B (Fig. 4). Nocardia cyriacigeorgica was the most abundant bacterial species in group B ( Table 1). The genus Nocardia comprises aerobic, Gram-positive, catalase-positive, and partial mycobacterial filamentous bacteria that exist in the soil and decaying plants and belong to the order Actinomycetes. Their transmission most commonly occurs via inhalation, and other potential routes include penetrating injuries, iatrogenic procedures, and food ingestion [44]. The risks for infection with Nocardia spp. include comorbidities such as diabetes, chronic lung disease, alcoholism, and active cancer, as well as immunocompromised conditions such as immunosuppressive therapy, and endovascular device placement [45][46][47]. Although an association with pre-existing diseases, such as diabetes, could not be verified in group B, the bacteria in this group were significantly correlated with low levels of serum albumin (p = 0.0026). The occurrence of different forms of immunodeficiency and malnutrition in individuals in group B may have been associated with the presence of N. cyriacigeorgica.
Kiyasu et al. [48] conducted a literature review of 30 infectious endocardial inflammation cases caused by Nocardia. It was reported that 12 patients (40%) among those examined were immunodeficient due to malignant tumors, organ transplantation, and malnutrition [48]. In addition, Nocardia was identified in the blood in 19 cases (63%), with blood vessels in the lungs being considered a route of entry [48]. In the subjects of our study, no findings suggestive of Nocardia infection were found in the skin and central nervous system [49], which are frequently infected organs including the lungs, and it was not possible to identify the route of entry into the blood. Additionally, Armingohar et al. analyzed bacterial DNA in vascular lesions, such as arteriosclerotic lesions, in the presence or absence of periodontal disease, reporting that intestinal bacteria were frequently detected along with oral bacteria in the patients with periodontal disease [50]. These results strongly suggest that swallowed oral bacteria and other bacterial species may reenter the systemic circulation from the intestinal tract. Thus, it is possible that the route of bacterial invasion via the intestinal tract is involved in the pathology of aortic stenosis, and we believe that further investigation is of paramount importance.
Recent epidemiological, clinical, and experimental analyses have supported the relationship between bacteremia or inflammation due to periodontal disease and systemic disease [51]. To the best of our knowledge, this study is the first to comprehensively characterize the composition of oral plaque and aortic valve microbiota in patients with aortic stenosis. Using metagenomic analysis, we identified bacterial species shared between the oral cavity and aortic valve tissues. Our results support that bacterial species that form the oral microbiota may translocate to the aortic valve due to bacteremia in aortic stenosis patients and thus may be associated with disease onset and severity. However, in this study, it was not possible to verify the mechanism by which bacteria attach to the aortic valve to affect the pathology of aortic stenosis. Additionally, owing to the high cost of metagenomic analysis ($830 per case) and occasional urgent need for surgery for aortic stenosis, only a limited number of samples from patients who provided consent were available. Despite this limitation, crucial events associated with the development of bacteremia in the pathology of aortic stenosis were adequately captured. It is sufficiently clear that the management of periodontal disease and proper oral care can positively affect the morbidity, mortality, and health care costs associated with nonoral systemic diseases. Further studies aimed at developing strategies for maintaining the integrity of the oral microbiota are warranted.

Conclusions
In this study, we investigated the relationship of oral microbiota dysbiosis and oral cavity conditions with the pathology and progression of aortic stenosis. We comprehensively characterized oral plaque and aortic valve microbiota in patients with aortic stenosis using metagenomic analysis.
Bacterial species, including oral bacteria, were detected in the aortic valve tissues. In addition, we found a correlation between the aortic valve microbiota and bleeding rate on probing. These results indicate that oral bacteria may reach the aortic valve via the bloodstream and cause inflammation and alter immune responses, thereby contributing to the development and progression of aortic stenosis. As this was a pilot study with a limited number of samples, we need to accumulate more cases and conduct further investigations to broaden our understanding of these relationships. Our study here provided new insights into the pathophysiology of aortic stenosis and suggests careful oral hygiene management of and prevention of systemic dissemination of oral bacteria might be a useful strategy for the prevention and treatment of aortic stenosis.