Microbial Communities and Functional Genes in Periodontitis and Healthy Controls

Abstract

Background

Periodontitis is a chronic progressive disease and the leading cause of tooth loss in adults. Recent studies have shown the impact of oral microbial communities on systemic health and diseases such as cancer, atherosclerosis, rheumatoid arthritis, inflammatory bowel disease, diabetes, hypertension, and Alzheimer's disease. In previous case control studies investigatin the relationship between periodontal disease and the oral microbiota, little attention has been paid to the intersections of these domains.

Methods

Here, we used high-throughput 16S rRNA sequencing to analyse the differences in the microbial composition in saliva between a group of patients with chronic periodontitis (C; n = 51) and a healthy control group (H; n = 61) and predicted the functional gene composition by Phylogenetic Investigation of Communities by Reconstruction of Unobserved States.

Results

We found significant alterations in oral microbial diversity between C and H (P = 0.002). Sixteen genera were significantly different between C and H, and 15 of them were enriched in C linear discriminant analysis (LDA > 2). Fifty functional genes were significantly different between C and H, and 34 of them were enriched in C (P < .025).

Conclusions

Periodontitis is associated with significant changes in the oral microbial community.

Introduction

Periodontitis is a chronic inflammatory disease caused by microorganisms.¹ It can lead to the loss of periodontal tissue adhesion and the loss of alveolar bone, which eventually leads to tooth loss.² The disease is the leading cause of tooth loss in adults, but it can also occur in children and adolescents.^3,4 In addition, periodontitis can increase the risk of cancer, atherosclerosis, and rheumatoid arthritis and affect the systemic immune response and glucose metabolism.^5,6

It is known that periodontitis is a polymicrobial infectious disease caused by potentially pathogenic microorganisms that interact synergistically.⁷ This pathogenic pattern indicates that any individual in the community may be susceptible to the disease, and even low-abundance strains may be key species that affect the behaviour of such complex communities.⁸ Therefore, the study of oral microbial communities and their dynamic diversity is critically important.

Human microbial communities contain a variety of microorganisms, which develop mutually beneficial relationships with the host under healthy conditions, but their dynamic balance is affected by the host and environmental factors.⁹ On the other hand, the host plays a significant role in the composition and gene expression of the host microbiota.^10,11 Changes in the oral environment disrupt the normal relationship between the host and its resident microorganisms, increasing the risk of disease.¹² Recent studies on the relationship between the oral microbiome and systemic diseases such as rheumatoid arthritis, inflammatory bowel disease, diabetes, hypertension, and Alzheimer's disease have shown that these systemic diseases may affect the oral microbiome.13, 14, 15, 16 Meanwhile, case control studies of oral microbiology and personal factors, including gender, age, obesity, dietary habits, smoking, and alcohol consumption, have shown that they can increase the risk of systemic or local disease and can further affect the composition of the oral microbiota.17, 18, 19, 20, 21 Wang et al²² found that there were significant differences in the community composition of oral bacteria after subgingival scaling, which gradually recovered and remained stable over time. Therefore, we need to consider factors regulating the composition of the oral microbiota through case control studies.

In this study, based on the exclusion of systemic, local, and confounding factors, we used 16S rRNA gene sequencing and Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) functional annotation to explore changes in the microbial taxonomic composition of the healthy periodontal population vs the periodontitis patient population to identify key microbial populations and to determine key genes and pathways associated with the pathogenicity of the periodontal microbial communities. Compared with existing studies, this study's purpose was to compare the microbial composition and functional genes of participants with only periodontitis with those of healthy controls no systemic diseases and oral diseases to elucidate the direct relationship between the oral microbiome and periodontitis.

Materials and methods

Experimental design and flow

We collected 386 saliva samples (representing 386 participants), and after excluding the systemic clinical diagnosis and exclusion procedures, 112 participants were included in this study (Figure 1A). Diagnostic criteria for hypertension included systolic blood pressure ≥140 mm Hg and/or diastolic blood pressure ≥90 mm Hg (≥3 times, on different days).²³ Diagnostic criteria for diabetes were glycosylated haemoglobin ≥6.4% (46 mmol/mol)²⁴ and so on. The 112 participants included 61 healthy controls (H) and 51 patients with chronic periodontitis (C). Finally, statistical analysis and bioinformatics analysis of the basic information and oral microbiome were performed, respectively. The study was approved by the Institutional Review Board of the First Affiliated Hospital of Zhengzhou University (ethics approval number is 2018-KY-90). All methods were carried out per relevant guidelines and regulations. All experimental protocols were approved by a named institutional and/or licensing committee. Informed consent was obtained from all participants and/or their legal guardian(s).

Fig. 1.

Oral microbial difference of operational taxonomic units (OTUs) distribution and phylum level. A, Study design and flow diagram. A total of 386 saliva samples were collected. After rigorous diagnosis and exclusion procedures, 61 H and 51 C were included. B, A Venn diagram showing the overlaps between C and H displayed that 1846 of the total richness of 2482 OTUs were shared, whilst 346 OTUs in C and 290 OTUs in H were unique. C, Composition of oral microbiota at the phylum level between the groups. The results showed that Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, and Fusobacteria were the top 5 bacteria in the richness of oral microflora at the phylum level between C and H，they had unchanged significantly microbial communities. C, chronic periodontitis; H, healthy control.

Participants

From June 2019 to December 2019, a total of 386 samples were obtained from participants recruited from the Health Management Center at the First Affiliated Hospital of Zhengzhou University, and participants meeting the following criteria were excluded from the study: hypertension; diabetes; oral mucosal lesions; dental caries; pericoronitis; metal porcelain restorations or removable dentures; invasive surgery, radiotherapy, and chemotherapy in the oral and maxillofacial region within 12 months; periodontal treatment within 12 months; and antibiotics or nonsteroidal anti-inflammatory drugs within 3 months. Inclusion criteria for the healthy controls (H) included the absence of gingival redness, bleeding on probing, attachment loss, and bone loss. Inclusion criteria (Supplementary Material, Table 1) for the chronic periodontitis group (C) were based on the 2018 International Classification of Periodontal Diseases.²⁵

Basic information and sample collection

Before sampling, basic information (Supplementary Material, Table 2) was collected, including gender, age, body mass index (BMI), dietary habits, smoking status, alcohol consumption, daily brushing frequency, daily flossing, and malocclusion. Unstimulated saliva (minimum 2 mL) with no sputum was collected. First, these samples were stored in a cryopreservation tube containing 2 mL of buffer solution and treated with an oscillator for 1 minute.²⁶ Then, we stored the samples at −80 °C.

Sample processing

DNA from saliva samples was extracted using the MagPure Swab DNA Kit (Magen) according to the manufacturer's recommendations. The V3 and V4 domain library sequences of 16S rRNA were constructed using 50 ng of DNA isolated from the sample. The V3 and V4 regions were amplified using polymerase chain reaction (PCR). Forward and reverse primers contained the sequences 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGR, RBGCASCAGKVRVGAAT-3′ and 5′- GTCTCGTGGGCTCGGAGATGTGTATA, and AGAGACAGGGACTACNVGGGTWTCTAATC-3′, respectively. PCR products were sequenced using the Illumina MiSeq platform at Precision Gene Inc.

Genomic DNA quantification

First, sequence splicing software FLASH²⁷ (Fast Length Adjustment of Short reads, v1.2.11) was used to assemble the pairs of reads sequenced from double ends into a sequence by using the overlap relationship, and the Tags in the highly variable region were obtained. The specific conditions are as follows: (1) The minimum matching length is 15 bp. (2) The allowable mismatch ratio of the overlap area is 0.1.

Then the software USEARCH²⁸ (v7.0.1090) was used to cluster the assembled Tags into operational taxonomic units (OTUs). The main process is as follows: (1) Using UPARSE to cluster at 97% similarity, the representative sequence of OTUs was obtained. (2) The chimerae generated by PCR amplification was removed from the OTU representative sequence using UCHIME²⁹ (v4.2.40). The chimeraes were removed by comparing with the existing chimerae database (gold database [v20110519]). (3) The usearch_global method was used to compare all Tags back to the OTU representative sequence to obtain the OTU abundance statistical table of each sample.

After obtaining the OTU representative sequence, the OTU representative sequence was combined with the database by RDP classifer (v2.2) software (Greengene V201305, RDP³⁰ Release11.5 2016-9-30; Silva V138 2019-12-16). Species annotation was performed for comparison, and the confidence threshold was set to 0.6. The comment results were filtered via the following process: (1) Remove OTU without comment results. (2) Unannotated results do not belong to the species in the analysis. The remaining OTUs can be used for later analysis.

Sequence data analysis

The QIIME toolkit was used for analysing the 16S rRNA data. The sequences were then divided into OTUs, representative sequences of OTUs were analysed for species taxonomy in conjunction with the Human Oral Microbiome Database (http://www.homd.org), and the community composition of each sample was counted at different species taxonomic levels. Based on the original data with 97% sequence identity, the quality-filtered reads were clustered to obtain 2482 OTUs in the data. We removed OTU sequences with fewer than 5 repetitions for quality control.

Statistical analysis and bioinformatics analysis

Statistical analyses and bioinformatics analyses were performed using R (version 4.0.2). Differences in continuous variables were detected using the independent-sample t test, and differences in categorical variables were detected using the chi-square test. Shannon index and Observed Species (Obs) index were used to show bacterial alpha diversity. Based on the Bray-Curtis distance matrix, principal coordinates analysis (PCoA) visualisation was performed to show the difference in beta diversity of microbial communities between cohorts. Linear discriminant analysis effect size (LEfSe) was used to search for statistically different biomarkers between cohorts. Spearman correlation analysis was conducted to analyse the correlation between the clinical index data of the participants and the relative abundance of the differential flora, and a heatmap was drawn to observe the relationship between the differential flora and its characteristics. PICRUSt was used to obtain the annotation results of the microbiome and gene function in the KEGG Orthology (KO) database, and the difference analysis was performed with a rank sum test to obtain the significant difference in gene function of the microbiome in different cohorts.

Results

Information on all participants

According to statistical analysis, there were no significant differences between the groups regarding gender, age, BMI, smoking status, alcohol consumption, daily brushing frequency, flossing, or malocclusion (P_all > .05; Table). There were 28 females and 23 males (mean age, 55.92 ± 8.58 years) with a BMI of 24.64 ± 6.62 kg/m² in the chronic periodontitis group (C). Meanwhile, there were 28 females and 33 males (mean age, 54.05 ± 6.84 years) with a BMI of 24.69 ± 4.66 kg/m² in the healthy control group (H).

Table.

Demographic characteristics of participants.

Characteristics		H (n = 61)	C (n = 51)	t/χ2	P value
Age	Years, mean ± SD	54.05 ± 6.84	55.92 ± 8.58	−1.285	.201a
Gender	Male	33	23	0.900	.343b
	Female	28	28
BMI	kg/m2, mean ± SD	24.69 ± 4.66	24.64 ± 6.62	0.048	.962a
DH	Never = 0	2	1	0.188	.910b
	Current = 1	28	24
	Former = 2	31	26
SS	Never = 0	47	40	0.492	.782b
	Current = 1	7	7
	Former = 2	7	4
AC	Never = 0	39	39	2.857	.240b
	Current = 1	13	9
	Former = 2	9	3
DBF	No = 0	41	38	0.712	.399b
	Yes = 1	20	13
DF	No = 0	60	51	0.016	.898b
	Yes =1	1	1
M	No = 0	44	37	0.002	.961b
	Yes = 1	17	14

AC, alcohol consumption; BMI, body mass index; C, chronic periodontitis; DBF, daily brushing frequency; DF, daily flossing; DH, dietary habit; H, healthy control; M, malocclusion; SS, smoking status.

^aDifferences in continuous variables were detected using the independent-sample t test.

^bDifferences in categorical variables were detected using the chi-square test.

Oral microbiome composition analysis based on 16S rRNA

The Venn diagram displayed that 1846 of the total richness of 2482 OTUs were shared between C and H, whilst 346 OTUs in C and 290 OTUs in H were unique to the dataset (Figure 1B), respectively. Based on the rank sum test, we performed a visual analysis of the richness of oral microbiota at the phylum level, and the results showed that Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, and Fusobacteria were the top 5 bacteria in the richness of oral microflora between C and H (Figure 1C).

Diversity analysis of the microbiotas at the genus level

Based on the Wilcoxon test, the Shannon index and Obs index showed that the community richness and evenness difference of C and H did not differ significantly at the genus level (P > .05; Figure 2A and B). PCoA under Bray-Curtis distance showed significant differences in the first principal components between C and H (P = .002, P = .83; Figure 2C, D, and E). The results showed that there were significant differences in microbial community diversity between C and H, suggesting that the oral microbiota changed significantly from periodontal health to periodontitis.

Fig. 2.

Diversity analysis, linear discriminant analysis effect size (LEfSe), and Spearman correlation analysis in bacterial communities at the genus level. A, B, Based on the Wilcoxon test, Shannon index, and observed species (obs) index showing communities’ richness and evenness difference of genus level between C and H (P > .05). C, D, Based on Bray-Curtis distance (P = .002, P = .83). E, Beta diversity was calculated using weighted UniFrac by principal coordinates analysis, indicating a distribution of oral microbial communities between C and H. F, LEfSe showing microbial genera with linear discriminant analysis (LDA) scores >2 and cladogram in C and H. G, Spearman correlation analysis showing the correlations between species’ abundances and participants’ characteristics in C and H. AC, alcohol consumption; BMI, body mass index; C, chronic periodontitis; DBF, daily brushing frequency; DF, daily flossing; DH, dietary habit; H, healthy control; M, malocclusion; SS, smoking status. *P < .05; **P < .01.

Differences and correlation analysis of the microbiotas at the genus level

LEfSe showed that 16 genera differed significantly between C and H, whilst 15 genera were enriched in C, including Bifidobacterium, Bacteroides, Barnesiella, Coprobacter, Parabacteroides, Tannerella, Lactobacillus, Lactococcus, Selenomonas, Streptobacillus, Treponema, and so on. Moreover, Haemophilus was enriched in H (LDA >2, Figure 2F). Spearman correlation analysis showed the correlations between species abundances and participants’ characteristics (Figure 2G). Flavonifractor, Lactobacillus, and Bifidobacterium were positively correlated with characteristics (malocclusion and daily flossing) and were enriched in C; Selenomonas, Treponema, and Haemophilus in C were negatively correlated with characteristics (gender and daily flossing); and Haemophilus in H was significantly negatively correlated with daily flossing.

Crucial microbial predicted functions related to periodontitis

Based on the KO database, we deduced its functional gene composition from the species composition obtained by 16S rRNA sequencing and analysed the functional differences between the cohorts. The results showed the top 50 functional genes with significant differences (P < .025, 2-sided test, Figure 3), 34 functional genes were enriched in C and 16 functional genes were enriched in H. Amongst the functional genes enriched in C, EC:2.7.13.3 (2-component system) is indirectly relevant to bacterial adaptation and colonisation and EC:6.3.1.2 (glutamine synthetase) is a basic part of nitrogen metabolism. Those associated with ATP synthesis were EC:5.4.2.12 (carbamoyl-phosphate synthase) and EC:3.6.3.14 (flagellum-specific ATP synthase); the functional genes related to glycometabolism were EC:4.1.1.3 (phosphoenolpyruvate carboxylase), EC:3.2.1.21 (beta-glucosidase), EC:2.7.11.1 (phosphorylase kinase gamma), and EC:2.7.1.11 (PFK/6-phosphofructokinase 1), respectively. Meanwhile, amongst the functional genes enriched in H, those genes associated with the immune system were EC:6.3.5.7 (glutamyl-tRNA [Gln] amidotransferase) and EC:6.3.5.6 (aspartyl-tRNA [Asn] amidotransferase); the functional gene associated with the ability to cross-link bacterial cell membranes was EC:6.3.2.4 (D-alanine-D-alanine ligase).

Fig. 3.

Functional gene variation in bacterial communities. Fifty functional genes differed significantly, and 34 were enriched in C (P < .025, 2-sided test). C, chronic periodontitis; H, healthy control.

Discussion

Periodontitis is an inflammatory host immune response caused by an imbalance in the bacterial composition of the oral cavity, resulting in microbial dysbiosis and chronic destructive inflammation. Excluding systemic and local confounding factors as much as possible, we investigated the differences in the microbial community composition in saliva samples between the chronic periodontitis (C) and the healthy control (H) groups and made functional predictions.

Griffen et al³¹ and Abusleme et al³² found that the alpha diversity of the oral microbial communities in C was significantly higher than that in H.^, In contrast, Chen et al³³ and Kirst et al³⁴ found no significant difference in the alpha diversity of the oral microbial communities between C and H. These studies are similar to the results of our study. In addition, Li et al³⁵ and Chen et al³⁶ found that the beta diversity of the oral microbial communities in C differed significantly from that in H. These studies are similar to the results of our study. These results indicated that the oral microbial community structure changes significantly from periodontal health status to periodontitis.

Many studies have been conducted to explore the composition of oral microbial communities and to describe the relationship between periodontitis and pathogenic microorganisms. Traditionally, A actinomycetemcomitans, P gingivalis, T forsythia, T denticola, F nucleatum, and P intermedia are considered to be the causative agents of periodontitis. In addition, Teles et al³⁷ confirmed that P intermedia, F nucleatum, C nodosum, and A actinomycetemcomitans are significantly associated with different periodontal states. Over the past decade, nonculture molecular biotechnology has found that Actinomyces, Prevotella, Streptococcus, Streptococcus, and Rothia were more abundant in C compared to H, whilst genera such as Megacoccus, Desulfovibrio, and Parvimonas were more abundant in H.³⁸ Chen et al³⁶ found that the saliva samples from C have relatively enriched in bacteria of the genera Eubacterium, Prevotella, Porphyromonas, Tannerella, Desulfobacterium, Methanobacterium, Bacteroidetes, and Bacillus difficile. In contrast, Chen et al³³ found that plaque samples from C were relatively enriched in bacteria of the genera Aeromonas, Denticola, Tannerella, Filifactor, and Aggregatibacter, whereas Streptococcus, Haemophilus, Aeromonas, Coccidioides, Campylobacter, and Bacillus were more abundant in H. In this study, 16 genera were found to be significantly different between C and H. Bacteria enriched in C included Bifidobacterium, Streptobacterium, Tannerella, Treponema, Bacteroides, Lactobacillus, Lactococcus, and Treponema. Meanwhile, Haemophilus was the only relatively abundant bacteria in H. In contrast to the findings of existing studies, the microorganisms that differed significantly in relative abundance between C and H in this study did not include Porphyromonas, which may be enriched in plaque samples from C. This may be attributed to the sample type. However, in our study, significantly different genera including Lactobacillus, Lactococcus, Treponema, and Leptospira are rarely found in existing studies and cannot be attributed to sampling type alone.

The 16S-based phylogenetic tree is very similar to clustering based on shared gene content,^39,40 and researchers often infer the properties of uncultured organisms from their cultured relatives. Thus, phylogeny and function were sufficiently linked, and this "predictive metagenomic" approach should provide useful insights into thousands of uncultured microbial communities where marker gene surveys and genomes are available.⁴¹ Functional gene array data on the composition and function of oral microbial communities for understanding human periodontal health and disease revealed a low number of major periodontitis genes but high signal intensity, with multiple virulence factors, amino acid metabolism, glycosaminoglycans, GAGs, and pyrimidine degradation–related genes enriched in periodontitis. This suggests their potential importance in periodontal pathogenesis. However, the relative abundance of genes involved in amino acid and pyrimidine synthesis was significantly lower compared to healthy controls.³⁵ In this study, functional analysis between these cohorts showed that the functional genes indirectly related to the immune system in C included EC:2.7.13.3 (2-component system).

In the application of 16s high-throughput sequencing, technical procedures used in different studies may cause the difference in sequencing results, such as amplified 16S rRNA regions (V1-V3 vs V3-V4), sequencing chemistry, read lengths, and sampling techniques.^34,42 Therefore, datasets across different studies could not be readily compared. Diverse microorganisms inhabit the oral cavity.³¹ Moreover, the microbial communities that grow on the different sites are also unique, including saliva, tongue, oral mucosa, mineralised tooth surfaces, and periodontal tissues.43, 44, 45, 46 The supragingival and subgingival dental plaque microbiome is evolutionally selected for different compositions, niche anatomy, antigen and immune exposure, and nutritional background.^47,48 Therefore, it is unsurprising that there are differences in the results from these studies as the criteria for collecting dental plaque samples are inconsistent. Whilst the dynamic changes of the microbial community in dental plaque biofilm fluctuated dramatically before and after clinical cleaning, the salivary microbial community remained relatively stable, which was the main source of dental plaque microbial supplement.²² With the technological advancement of detection, more microbes, proteins, and inorganic ions can be found in saliva samples.⁴⁹ Saliva plays an important role in regulating the oral microbiome and maintaining oral health.50, 51, 52 This provides a target for detecting the changes in the oral microenvironment. Saliva is relatively easier to collect compared to plaque. Hence, saliva is particularly suitable as a sample of oral microbes to monitor changes in the oral microenvironment. In addition, we excluded participants who had low salivation due to local or systemic factors.

Therefore, the above factors have significant impacts on the oral microbiome. Previous cohort studies on periodontitis and oral microbiome excluded some of the above confounding factors. In addition, correlation analysis of other remaining factors was not conducted, so the results obtained may have certain limitations. We excluded the currently recognised influencing factors as far as possible in the study, and multifactor analyses were peroformed to improve the authenticity and reliability of the data.

However, this study also has some shortcomings. For example, the sample number and the type could be further increased. Follow-up studies should address these shortcomings.

Conclusions

This study showed that microbial communities and functional genes significantly differed between saliva samples from healthy controls and those from patients with chronic periodontitis.

Conflict of interest

None disclosed.

Appendix. Supplementary materials

Supplementary Table 1. The inclusion criteria of these subjects.

Supplementary Table 2. The basic information of these subjects.

Supplementary Table 3. Genus.

Supplementary Table 4. Function.