Periodontitis and type 2 diabetes are connected pandemic diseases, and both are risk factors for cardiovascular complications. Nevertheless, the molecular factors relating these two chronic pathologies are poorly understood. We have shown that, in response to a long-term fat-enriched diet, mice present particular gut microbiota profiles related to three metabolic phenotypes: diabetic-resistant (DR), intermediate (Inter), and diabetic-sensitive (DS). Moreover, many studies suggest that a dysbiosis of periodontal microbiota could be associated with the incidence of metabolic and cardiac diseases. We investigated whether periodontitis together with the periodontal microbiota may also be associated with these different cardiometabolic phenotypes. We report that the severity of glucose intolerance is related to the severity of periodontitis and cardiac disorders. In detail, alveolar bone loss was more accentuated in DS than Inter, DR, and normal chow-fed mice. Molecular markers of periodontal inflammation, such as TNF-α and plasminogen activator inhibitor-1 mRNA levels, correlated positively with both alveolar bone loss and glycemic index. Furthermore, the periodontal microbiota of DR mice was dominated by the Streptococcaceae family of the phylum Firmicutes, whereas the periodontal microbiota of DS mice was characterized by increased Porphyromonadaceae and Prevotellaceae families. Moreover, in DS mice the periodontal microbiota was indicated by an abundance of the genera Prevotella and Tannerella, which are major periodontal pathogens. PICRUSt analysis of the periodontal microbiome highlighted that prenyltransferase pathways follow the cardiometabolic adaptation to a high-fat diet. Finally, DS mice displayed a worse cardiac phenotype, percentage of fractional shortening, heart rhythm, and left ventricle weight-to-tibia length ratio than Inter and DR mice. Together, our data show that periodontitis combined with particular periodontal microbiota and microbiome is associated with metabolic adaptation to a high-fat diet related to the severity of cardiometabolic alteration.
- periodontal dysbiosis
- periodontal microbiome
- metabolic diseases
- high-fat diet
the oral cavity is one of the most dynamic biotypes for human bacteria. It houses a complex buccal microbiota composed of >1010 bacteria separated into >450 species in adulthood (26). A number of oral microbiota linked to various oral niches, such as tooth surface, saliva, internal jowl, tonsils, and periodontal tissue, have been described (18). Notably, Actinomyces, Prevotella, Streptococcus, and Fusobacterium are the most common genera in periodontal microbiota (24), and saliva presents a higher bacterial diversity than periodontal tissue and a microbial community different from that of periodontal tissue (28). This diversity induces a large variety of immune factors that contribute to healthy symbiosis between the oral microbiota and oral tissue (10). A break (dysbiosis) in this equilibrium could target the development of oral diseases such as periodontitis, an oral chronic infection of the periodontium caused by an increase in gram-negative anaerobic bacteria in the periodontal microbiota (6). This infection promotes a local inflammation that induces a periodontal bone loss. Hence, an oral microbiota dysbiosis with an increased ratio of gram-negative to gram-positive bacteria (39) may be considered a risk factor of metabolic diseases, as in the case of dysbiosis of the gut microbiota (41). The prevalence of periodontitis is 20–50% in the general population (35); however, it is considered the sixth most commonly occurring complication of diabetic patients, with a prevalence of 60%. Since epidemiological studies predict 500 million diabetic patients in 2030 (2), 300 million could develop periodontitis. The origin of this accelerating development is related to several interacting factors such as sedentary lifestyle and unhealthy eating habits. Interestingly, different metabolic phenotypes from diabetic-sensitive (DS) to -resistant (DR) can be identified in a mouse model of nutritional stress, suggesting a metabolic adaptation to a high-fat diet (HFD) (11). Furthermore, the appearance and the severity of complications are linked to the severity of diabetic phenotype in patients (3). Although the prevalence of cardiovascular disease (CVD) is higher in diabetic patients, the link cannot be explained by traditional CVD risk factors such as fasting hyperglycemia, obesity, dyslipidemia, and hypertension (2). Recently, it has been demonstrated that the incidence of metabolic disease and CVD (8) is also associated with dysbiosis of the gut microbiota (12). In this context, several studies have identified oral infections as new comorbidities for CVD (26), but the strong epidemiological association between CVD, diabetes, and periodontitis lacks a clear biological explanation. However, many pathogens of periodontitis have been implicated in metabolic (9) and systemic diseases such as CVD (26).
Dysbiosis of the periodontal microbiota could be associated with the incidence of metabolic (12) and cardiac (7) diseases. To demonstrate this link, we fed a cohort of 100 mice a diabetogenic/nonobesogenic HFD (11) for 3 mo and analyzed the features of periodontitis, the periodontal microbiota and microbiome, and glucose and cardiac metabolism. We report that periodontitis and dysbiosis of the periodontal microbiota are distinguishing traits of metabolic adaptation to a low-carbohydrate HFD in association with altered cardiac metabolism.
Animals and experimental procedures.
C57BL/6J wild-type male mice (Charles River, L'Arbresle, France) were group-housed (6 mice per cage) in a specific pathogen-free controlled environment (inverted 12:12-h daylight cycle, lights off at 10 AM). At 5 wk of age, mice were fed a diabetogenic, high-fat carbohydrate-free diet [HFD; energy content: 72% fat (corn oil and lard), 28% protein, and <1% carbohydrate; SAFE, Augy, France] (42). After 3 mo, mice were characterized according to glucose tolerance as follows: DR (glycemic index <5,500 mg·dl−1·min−1), intermediate (Inter; glycemic index 7,000–8,000 mg·dl−1·min−1), and DS (glycemic index >8,500 mg·dl−1·min−1) mice. A control group was fed a normal chow (NC) diet. All animal experimental procedures were approved by the local ethical committee of Rangueil University Hospital (Toulouse, France).
Intraperitoneal glucose tolerance test.
Glucose (1 g/kg) was injected into the peritoneal cavity of 6-h-fasted mice. Blood glucose in 2 μl of blood collected from the tip of the tail vein was measured with a glucometer (Roche Diagnostics, Meylan, France) at −30, 0, 30, 60, 90, and 120 min after the glucose injection.
Quantification of mandibular alveolar bone loss.
Hemimandibles were scanned using a high-resolution μCT (Viva CT40, Scanco Medical, Bassersdorf, Switzerland) (43). Data were acquired at 45 keV, with a 10-μm isotropic voxel size. Six linear measurements were obtained from each molar (total of 3 molars per mandible) via a stereomicroscope with an on-screen computer-aided measurement package. The alveolar bone loss (ABL, in mm) was measured from the cementoenamel junction to the alveolar bone crest for each molar.
Real-time quantitative PCR analysis.
Total RNA from periodontal tissue was extracted using the TriPure reagent (Roche, Basel, Switzerland). cDNA was synthesized using a reverse transcriptase (Applied Biosystems, Foster City, CA) from 1 μg of total RNA as previously reported (9). Expression of common genes of inflammation was analyzed with the following primers (5′ to 3′; Eurogentec, San Diego, CA): TGGGACAGTGACCTGGACTGT (forward) and TCGGAAAGCCCATTTGAGT (reverse) for tumor necrosis factor-α (TNF-α), TCGCTCAGGGTCACAAGAAA (forward) and CATCAGAGGCAAGGAGGAAAAC (reverse) for interleukin (IL)-1β (IL-1β), ACAGCCTTTGTCATCTCAGCC (forward) and CCGAACCACAAAGAGAAAGGA (reverse) for plasminogen activator inhibitor-1 (PAI-1), and ACAAGTCGGAGGCTTAATTACACAT (forward) and TTGCCATTGCACAACTCTTTTC (reverse) for IL-6. The concentration of each mRNA was normalized for RNA loading against the ribosomal protein L19 [5′ to 3′: GAAGGTCAAAGGGAATGTGTTCA (forward) and CCTTGTCTGCCTTCAGCTTGT (reverse)] as an internal standard, and the data were analyzed according to the cycle threshold (2−ΔΔCT) method.
Periodontal microbiota analysis.
Total DNA was extracted from frozen mandibles as previously described (42). The whole 16S bacterial DNA V2 region was targeted by the 28F-519R primers and analyzed using the FLX pyrosequencer [454 Life Sciences at Roche Technologies Research & Testing Laboratory (http://www.researchandtesting.com/)]. An average of 4,907 sequences were generated per sample.
Echocardiography was carried out on lightly anesthetized (1% isoflurane in air) mice placed on a heating pad. Left ventricle function and heart rate were obtained during TM mode acquisition from the short-axis view at the level of the papillary muscles using a Vivid7 echograph and a 14-MHz transducer (i3L, GEHealthcare). Images were transferred and analyzed off-line with EchoPAC (GEHealthcare). Echocardiography was performed with protocols blinded with respect to the different groups. Mice were lethal-anesthetized with pentobarbital, heart weight and left ventricle weight (LVW) were determined, and tibia length (TL) was measured to calculate the cardiac hypertrophy index (LVW-to-TL ratio) (20).
Values are means ± SE. One-way ANOVA followed by Tukey's posttest was used to assess intergroup differences, except for the intraperitoneal glucose tolerance test (IPGTT), where two-way ANOVA followed by Bonferroni's posttest was applied. P < 0.05 defined statistical significance. Statistical analyses were performed using GraphPad Prism version 5.00 for Windows Vista (GraphPad Software, San Diego, CA). The cladogram and linear discriminant analysis score on Fig. 3 were drawn by the Huttenhower Galaxy website (http://huttenhower.sph.harvard.edu/galaxy/) via the LEfSe algorithm (44); linear discriminant analysis in Fig. 6 was drawn using XLSTAT for Microsoft Windows Excel.
Metabolic adaptation to a HFD results in mice with three metabolic phenotypes: DR, Inter, and DS.
We previously described an animal model of metabolic adaptation to a diabetogenic/nonobesogenic low-carbohydrate HFD driving different metabolic phenotypes in a cohort of mice (42). By evaluating glucose tolerance during an IPGTT and body weight, we validated three different phenotypes of mice, DR, Inter, and DS (Fig. 1, A–D), after 3 mo of HFD feeding. In addition, body weight was not significantly affected by this metabolic adaptation to a HFD, as previously shown (42), but was significantly reduced compared with NC-fed mice (Fig. 1C).
Periodontitis and dysbiosis of the periodontal microbiota are traits associated with metabolic adaptation to a HFD.
We previously showed that a HFD induces periodontitis in mice through the CD14 LPS receptor signaling (9). Here, we investigated whether the severity of metabolic alteration could be associated with the severity of periodontitis. ABL was higher in all HFD- than NC-fed mice, but ABL increased from DR to Inter to DS mice (Fig. 2A). RNA expression of TNF-α, PAI-1, and IL-1 was higher mainly in DS mice, indicating an increased periodontal inflammation (Fig. 2B). Additionally, TNF-α and PAI-1 expression in periodontal tissue were positively and significantly correlated with ABL (Fig. 2, C and D) and glycemic index (Fig. 2, E and F).
To evaluate whether the association between metabolic phenotype and periodontitis may also be linked to a dysbiosis of the periodontal microbiota, we profiled the periodontal microbiota in all groups of mice. The phylum Actinobacteria was significantly elevated in NC-fed mice compared with the other groups, whereas the class Cyanobacteria was significantly increased in the Inter phenotype, as shown by the cladogram based on linear discriminant analysis effect size (Fig. 3, A and B). We also showed, in detail (phyla to species), changes in the periodontal microbiota for all groups (Fig. 3, C–O). The periodontal microbiota of DR mice was dominated by the phylum Firmicutes (Fig. 3C) and the family Streptococcaceae (Fig. 3F). By contrast, the families Porphyromonadaceae and Prevotellaceae showed a trend to be lower in DR mice (Fig. 3, G and H). Tannerella and Prevotella, major periodontal pathogens, were not detected in DR mice compared with Inter and DS mice at the level of genus (Fig. 3, J and K) and species (Fig. 3, M and N) but seemed to be higher (although nonstatistically significant) in NC mice (Fig. 3, M and N). In addition, the level of Streptococcus species tends to be higher in DR mice than in the other groups (Fig. 3O). Furthermore, the periodontal microbiota of DR mice was characterized by a lower diversity from phylum to species (Fig. 3, C–P), as underlined in Fig. 3P. Together, these data demonstrate that periodontitis and dysbiosis of the periodontal microbiota may characterize metabolic adaptation to the low-carbohydrate HFD.
Periodontitis and the periodontal microbiome follow metabolic adaptation to a HFD.
Next, we analyzed the periodontal microbiota from a functional point of view by performing a PICRUSt-based analysis of the periodontal microbiome. Two major pathways stand out to be potentially modulated in relation to the diabetic phenotype. 1) Retinoic acid-inducible gene (RIG)-I-like receptor signaling [implicated in the expression of inflammatory cytokines (i.e., TNF-α) and epithelial cytokines (i.e., IL-8) in response to microbial antigens] was significantly downregulated in DR and DS compared with Inter and NC mice (Fig. 4A). 2) The prenyltransferase pathway, regulating sterol synthesis and protein relocalization, was downregulated from NC to DR, Inter, and DS mice (Fig. 4B). The proportion of sequences linked to the prenyltransferase pathway was significantly and negatively correlated with ABL (Fig. 4C), TNF-α periodontal expression (Fig. 4D), and glycemic index (Fig. 4E).
These data suggest that some functional pathways of the periodontal microbiome may be associated with metabolic adaptation to a HFD in mice.
Metabolic adaptation, together with periodontal dysbiosis, is also associated with the severity of cardiac alterations.
We previously reported that periodontitis can target cardiac metabolism in mice (7). Here we show that the three phenotypes obtained under HFD can be distinguished according to the ratio of LVW to TL (Fig. 5A), a standard index of cardiac hypertrophy, and with the association between this parameter and the glycemic index (Fig. 5B). Moreover, the percentage of fractional shortening, together with heart rhythm, was worsened in DS compared with DR and Inter mice (Fig. 5, C and D). Interestingly, the prenyltransferase pathway was significantly and positively correlated with the ratio of LVW to TL (Fig. 5E). Overall, NC-fed mice displayed a cardiac phenotype close to that of DR mice (Fig. 5. A–D).
Finally, using principal component analysis, we determined whether clusters of parameters may be found between bacterial family abundance in the periodontal microbiota and cardiometabolic parameters such as body weight, glycemic index, percentage of fractional shortening, heart rhythm, ABL, heart weight, heart weight-to-body weight ratio, TL, heart weight-to-TL ratio, LVW-to-TL ratio, and proinflammatory cytokines in periodontal tissue in all groups of mice. This analysis shows a complete separation of all groups, with a slight intersection between DS and Inter (Fig. 6), confirming the association between all the cardiometabolic parameters analyzed and metabolic adaptation to the HFD.
Collectively, these data suggest that metabolic phenotypes are correlated to periodontitis, dysbiosis of the periodontal microbiota, and the microbiome, together with cardiac alterations.
This study shows that the severity of periodontitis, together with dysbiosis of the periodontal microbiota, is associated with cardiometabolic adaptation to a HFD in mice. Actually, ABL and periodontal tissue inflammation, two key features of periodontitis, gradually correlated with the impairment of glucose metabolism, from the DR phenotype to the Inter and DS phenotypes. In this context, beyond the degenerative multiorgan complications of diabetes (32), diabetic patients are prone to develop periodontitis with an increased severity (35). Indeed, we show that the periodontal microbiota of DS mice is characterized by the families Porphyromonadaceae and Prevotellaceae, two aggressive bacteria in periodontal pockets (17). However, these two families were apparently increased (nonstatistically significant) also in NC-fed mice, although without periodontitis, because periopathogens can develop major periodontal impacts with metabolic consequences when on a HFD context, as we recently showed (5).
Interestingly, Prevotella intermedia has a higher prevalence in diabetic patients (46), and the decrease in the family Prevotellaceae upon partial deletion of gut microbiota by antibiotic treatment increased the incidence of diabetes in an insulin-dependent diabetes mouse model (13). In accordance with our data, species from Streptococcus characterize healthy periodontal microbiota compared with the periodontitis microbiota in subjects without systemic disease (46). By contrast, species from Tannerella, aggressive periodontal pathogens, are dominant in the periodontal microbiota of diabetic patients (46). Thus the decreased abundance of Streptococcus species and the increase in gram-negative periodontal pathogens may be considered as risk factors for the development of periodontitis, as suggested by our data. These families (Tannerella and Porphyromonadaceae) of gram-negative LPS-releasing periodontal pathogens are able to induce an inflammatory immune response involved in the development of periodontitis (39). In this context, both gut microbiota and complement are required for Porphyromonas gingivalis (Pg)-induced inflammation of the periodontal tissue (22), which may arise from degradation of periodontal connective tissue, as shown in mice (23). Pg is one of the major periopathogens (5, 19, 37), and we recently showed that Pg-induced periodontitis also induces a periodontal dysbiosis and insulin resistance by impairing the adaptive immune system (5). Hence, our data confirm the need for a correct immune program to manage periodontal pathogen infections (22). Among these, bacteria from the family Porphyromonadaceae are associated with CVD (38) and detected in atherosclerotic plaques (29), and one main inducer of periodontitis is Pg (33). We recently showed that Pg-induced periodontitis drives periodontal dysbiosis and promotes insulin resistance by dampening the adaptive immune system (5). Interestingly, Pg-induced effects on the periodontal tissue are both gut microbiota- and complement-dependent (22) and may arise from degradation of periodontal connective tissue, which occurs to a lesser extent in axenic than conventional mice (23). Finally, the systemic effects of Pg could originate from an altered gut barrier, leading to increased gut permeability and spread of bacteria into the systemic circulation (33).
Importantly, the family Porphyromonadaceae is associated with chronic diseases such as CVD (38) or nonalcoholic steatohepatitis (19) and with higher prediabetes prevalence among nondiabetic adults (45). In our model, cardiac functions also underwent metabolic adaptation following metabolic glucose impairment, since from NC-fed to DR to DS mice, we could observe a reduced percentage of fractional shortening and an increase in heart rhythm. Thus the occurrence of periodontal pathogens in diabetic patients could be targeted for the management of cardiac parameters (36). This strategy may also be applied to the intensive glucose control, which alone has harmful effects [increased mortality without affecting cardiovascular events (1)], beyond expected beneficial effects on both major macro- and microvascular events, albeit related to reduced nephropathy (1a). Therefore, it may be possible that analyses of the periodontal and gut microbiota may be useful to explain the above-reported divergence for two very close interventions on humans.
With regard to PICRUSt analysis of the periodontal microbiome, two pathways were potentially modulated in our model of metabolic adaptation to a HFD. The RIG-I-like receptor signaling pathway is implicated in prokaryotic DNA recognition (25). Hence, the concomitant production of type 1 IFN, inflammatory cytokines, and chemokines may lead to maturation of dendritic cells linked to periodontal inflammation and may, at least for Inter mice, explain the increased inflammation observed in this phenotype (Fig. 2B). This inflammation may be linked to microbial antigens released in periodontal tissue (39). Similar to the profile of the periodontal microbiota, the RIG-I-like receptor signaling pathway did not follow the metabolic adaptation to the HFD, since it was increased in NC-fed mice. The discordance observed for the RIG-I-like receptor signaling pathway in NC-fed mice is likely due to the lack of inflammation in these mice (Fig. 2B).
By contrast, the prenyltransferase pathway strictly follows the metabolic adaptation from the NC to the DS phenotype; this pathway is one of the key enzymes in hopanoid metabolism and is responsible for the stability of bacterial membranes (30). Thus we can hypothesize that modulations of the prenyltransferase pathway and the associated membrane instability could be linked to the production of bacterial metafactors in the periodontal environment, leading to increased inflammation (16), as the one characterizing metabolic diseases (8). Moreover, we previously showed that close metabolic phenotypes (i.e., glucose-tolerant mice) may be associated with a different gut microbiota (42). Therefore, we may also suggest that two different profiles of periodontal microbiota can be associated with close metabolic phenotypes such as DR and NC-fed mice.
Interestingly, in DS mice the general profile at the phylum level of the periodontal microbiota is strictly related to one of the gut microbiota that we observed during metabolic adaptation to a HFD and characterized by decreased Firmicutes-to-Bacteroidetes ratio (42). This profile is likely to increase the proportion of gram-negative bacteria, producing inflammatory bacterial antigens such as LPS, shown to initiate metabolic diseases (14). Moreover, a microbiota dominated by gram-negative bacteria may directly target metabolically active organs such as muscle, adipose tissue, and liver (2), aggravating metabolic diseases (15). Finally, the elevated LPS plasma levels in diabetic patients could be linked to periodontitis (40), supporting the bidirectional relationship between these two pathologies (31). We also showed that diversity of the periodontal microbiota increased during periodontitis, confirming our article (5).
V. Blasco-Baque was supported by the French Society of Arterial Hypertension (Société Française d'HyperTension Artérielle) and the French Diabetes Society (Société Francophone du Diabète).
No conflicts of interest, financial or otherwise, are declared by the authors.
M.B., F.R., P.L., P.M., A.W., V.A., A.C., A.G., M.S., and V.B.-B. performed the experiments; R.P., J.S.I., F.T., C.H., M.S., and V.B.-B. analyzed the data; F.T., M.S., and V.B.-B. edited and revised the manuscript; C.H., M.S., and V.B.-B. interpreted the results of the experiments; R.B., M.S., and V.B.-B. developed the concept and designed the research; M.S. and V.B.-B. prepared the figures; M.S. and V.B.-B. drafted the manuscript; M.S. and V.B.-B. approved the final version of the manuscript.
We thank Dr. Luc Malaval (INSERM U1059) for technical support, Lorette Gaffié and Dr. Robert Cameron for editing the English, and the animal facility of Rangueil INSERM/UPS US006 CREFRE and the microtomography facility in the medical faculty of the University Jean Monnet (St. Etienne, France).
- Copyright © 2016 the American Physiological Society