Reviews

Local and systemic mechanisms linking and inflammatory comorbidities

George Hajishengallis 1 ✉ and Triantafyllos Chavakis 2 Abstract | Periodontitis, a major inflammatory disease of the oral mucosa, is epidemiologically associated with other chronic inflammation-driven disorders, including cardio-metabolic, neurodegenerative and autoimmune diseases and cancer. Emerging evidence from interventional studies indicates that local treatment of periodontitis ameliorates surrogate markers of comorbid conditions. The potential causal link between periodontitis and its comorbidities is further strengthened by recent experimental animal studies establishing biologically plausible and clinically consistent mechanisms whereby periodontitis could initiate or aggravate a comorbid condition. This multi-faceted ‘mechanistic causality’ aspect of the link between periodontitis and comorbidities is the focus of this Review. Understanding how certain extra-oral pathologies are affected by disseminated periodontal pathogens and periodontitis-associated systemic inflammation, including adaptation of bone marrow haematopoietic progenitors, may provide new therapeutic options to reduce the risk of periodontitis-associated comorbidities.

17 (Box 2) Inflammatory bowel disease Periodontitis is a chronic inflammatory disease that pro- memory suggest that the inflammatory adapta- (IBD). A chronic inflammatory gressively affects the integrity of the tissues supporting tion of haematopoietic progenitors in the bone marrow disorder of the gastrointestinal the teeth (Box 1) and is associated epidemiologically with may connect multiple chronic inflammatory diseases18. tract with a complex several chronic disorders, such as cardiovascular disease, The mechanisms linking periodontitis to extra-oral aetiology, including genetic, environmental, microbial type 2 diabetes mellitus (T2DM), rheumatoid arthritis, comorbidities are consistent with clinical observations and host immune factors. inflammatory bowel disease (IBD), Alzheimer disease, associating periodontitis with bacteraemias, low-grade Two major IBD conditions nonalcoholic fatty liver disease and certain cancers1–5 systemic inflammation, increased myelopoietic activity are Crohn’s disease (which (Fig. 1). Whereas an imbalanced interaction between the and the ability of local periodontal treatment to attenuate may affect anywhere in the periodontal microbiome/microbiota and the inflamma- systemic inflammatory markers and improve comorbid digestive tract) and ulcerative 1,19–22 colitis (which affects only tory response of the host can readily explain local tissue disease activity (surrogate markers) . Conversely, the colon). destruction in periodontitis (Box 1), it has been largely systemic diseases such as T2DM can promote suscep- uncertain whether and how this imbalanced relationship tibility to periodontitis by increasing the inflammatory can causally link periodontitis and extra-oral comor- burden on the or by modulating the per- bidities. From a medical and therapeutic standpoint, iodontal microbiome1,23,24. Therefore, although a recip- it is imperative to understand whether the link between rocal relationship likely exists between periodontitis and periodontitis and associated comorbidities is merely a comorbidities, this Review focuses primarily on how correlative one or orchestrated by causal mechanistic periodontitis influences comorbid conditions. interactions. The Review aims to substantiate that the mouth– 1Department of Basic and Translational Sciences, Penn Recent animal model studies have demonstrated body link is not simply a consequence of common risk Dental Medicine, University biologically plausible mechanisms whereby periodonti- factors but is driven, in substantial part, by multiple of Pennsylvania, Philadelphia, tis may increase susceptibility to comorbidities6–16. Here, microorganism-induced immunological mechanisms PA, USA. we discuss how periodontal or locally activated that converge to increase the susceptibility of patients 2Institute for Clinical lymphocytes may disseminate to extra-oral tissues, with periodontitis to non-communicable chronic Chemistry and Laboratory where they may cause inflammatory and functional inflammatory diseases. Although it is not possible to Medicine, Faculty of Medicine, 25 Technische Universität complications (for example, endothelial cell dysfunction, cover all potential comorbid associations , emphasis Dresden, Dresden, Germany. bone marrow alterations, gut dysbiosis and immune sup- is given to epidemiologically strong associations ✉e-mail: [email protected] pression) that aggravate or in certain cases even initiate that are also supported mechanistically by clinically 6–16 https://doi.org/10.1038/ comorbid pathologies . Moreover, recent advances in relevant experimental evidence. Achieving a compre- s41577-020-00488-6 understanding the epigenetic basis of innate immune hensive mechanistic understanding of periodontal

426 | July 2021 | volume 21 www.nature.com/nri Reviews

Box 1 | Overview of periodontitis and its animal models Periodontitis became prevalent during the transition from hunter–gatherer to Neolithic farming societies Microbiome/microbiota (∼10,000 years ago), coinciding with dietary changes and Periodontal Microbiota is a diverse a compositional shift of the oral microbiota to a community bone loss microbial community that enriched in periodontitis-associated bacteria, including exists within a defined Porphyromonas gingivalis149. This oral disease remains a major anatomical niche (for example, public health and socioeconomic burden and afflicts, in its an environmentally exposed severe form, 10% of adults150. Periodontitis is an exemplar surface of a mammalian ∼ of an imbalanced interaction between the local microbiome organism). The term microbiome represents the and the inflammatory response of the host (dysbiosis, see microbial community, its the figure). Although induced ostensibly to control the combined genetic material microbial challenge, the inflammatory response is ineffective and its collective functions. and poorly controlled in susceptible individuals, leading Inflammation to inflammatory destruction of the periodontium, a Dysbiosis collective term for the tissues that surround and support the An imbalanced interaction, dentition (gingiva, periodontal ligament and alveolar bone)87. Complement–TLR Tissue breakdown between bacteria in a If untreated, the disease can lead to tooth loss, compromising cross-talk products (nutrients) community and/or between mastication, aesthetics and quality of life150. Periodontitis the microbial community and the host immune system, can be simulated in animals by placing silk ligatures which has detrimental effects around posterior teeth (ligature-induced periodontitis); on the host (as in periodontitis this generates a -retentive milieu leading to selective or inflammatory bowel expansion of indigenous pathobionts and induction of disease). The microbial gingival inflammation and bone loss151,152. Another widely imbalance derives from used model involves oral gavage with human periodontal changes in the abundance pathogens, most often using P. gingivalis. This pathogen and/or influence of individual colonizes the periodontium and causes quantitative and species relative to their compositional alterations to the commensal microbiota, Dysbiotic microbiome abundance or influence 153,154 in health. thereby instigating inflammatory bone loss . Bone loss is mediated — in both mice and humans — largely by T helper 17 (TH17) cell-derived IL-17, which induces expression of Periodontal treatment receptor activator of NF-κB ligand (RANKL) in osteoblasts, leading to osteoclast activation and bone resorption60,151. A procedure that involves The pathogenic process is not linear but instead represents a positive-feedback loop between the dysbiotic microbiome mechanical and the host inflammatory response. Indeed, while dysbiosis enhances destructive inflammation (largely through activation () to of complement and Toll-like receptors (TLRs)), inflammation generates a nutritionally favourable environment for selective remove the microbial biofilm expansion of periodontitis-associated organisms, termed inflammophilic pathobionts. This reciprocal reinforcement () and between dysbiosis and inflammation perpetuates a vicious cycle that drives periodontal disease pathogenesis87. (tartar) from the tooth surfaces and beneath the gingiva to enable inflammation resolution, comorbidities may not only support the notion that per- Periodontitis-associated systemic inflammation likely and to smooth the root surfaces to deter further iodontitis is a modifiable risk factor for life-threatening results from haematogenous dissemination of periodon- biofilm/calculus build-up. disorders but may also lead to new overarching concepts tal bacteria or spillover of inflammatory mediators from Advanced periodontitis explaining comorbid inflammatory pathologies, thereby periodontal tissues to the bloodstream. In this regard, may additionally require revealing new therapeutic options applicable to multiple the ulcerated epithelium of the periodontal pockets covers periodontal surgery, such inflammatory disorders. a surface area of 8–20 cm2 and may allow bacteria and as , to reduce the depth of the their products (for example, lipopolysaccharide (LPS) pockets between the teeth Periodontitis and systemic inflammation or proteases) to reach the circulation, causing bacter- and the surrounding gingiva. Although periodontitis shares inflammatory effector aemias (Fig. 2), which are documented in patients with mechanisms, as well as genetic and acquired risk factors, periodontitis5,21. Inflammation in extra-oral sites can be Inflammophilic From inflammation and the with many comorbid conditions, an independent asso- induced also through oro-pharyngeal or oro-digestive Greek suffix philic indicating ciation still remains between periodontitis and comor- translocation of periodontal bacteria; the former is asso- fondness. A property of bidities even after adjustment for confounders1,26,27. ciated with aspiration pneumonia34, whereas the latter bacteria that thrive under A possible factor contributing to this independent is associated with intestinal dysbiosis and gut-mediated inflammatory conditions by association is that periodontitis can cause low-grade systemic inflammation6,8,35,36. The elevated systemic utilizing products derived from the inflammatory breakdown systemic inflammation, which may influence the deve­ inflammation associated with periodontitis may have of tissues as nutritional lopment of comorbidities. In comparison with healthy multiple systemic complications as outlined below. substrates. controls, patients with severe periodontitis have elevated levels of pro-inflammatory mediators (such as IL-1, IL-6, Immunometabolic alterations Periodontal pockets The physiological narrow C-reactive protein (CRP) and fibrinogen) and increased Evolutionarily, host survival was strongly dependent 5,28–30 space between the tooth root neutrophil numbers in the blood . A prospective on two traits: a powerful immune response to fight off and the free gingiva is known study involving 11,869 individuals showed that failure infections and an effective calorie-storage system to as subgingival crevice; during to maintain was associated with elevated withstand starvation. Although the haematopoietic/ periodontitis progression, low-grade systemic inflammation and enhanced risk immune system and liver and adipose tissue represent however, this crevice deepens 31 into a periodontal pocket, of cardiovascular disease (CVD) . Conversely, suc- distinct tissues and organs in higher organisms, their which is a pathognomonic cessful local periodontal treatment attenuates systemic functions in primitive organisms are performed by a feature of the disease. inflammatory markers5,19,22,29,30,32,33. single organ (for example, the fat body in Drosophila)37.

NATuRe RevIewS | IMMunOlOgy volume 21 | July 2021 | 427 Reviews

Nonalcoholic fatty liver Not surprisingly, the host immune response is tightly A recent study showed that effective periodontal disease/nonalcoholic intertwined with metabolism, and dysfunction of this treatment reduces systemic inflammatory markers and steatohepatitis integrated system may lead to chronic metabolic– favourably influences the metabolic status of patients (NAFLD/NASH). NAFLD inflammatory disorders, such as obesity and related with T2DM (who showed reduction of plasma glucose is a condition associated HbA with obesity and metabolic pathologies, including the metabolic syndrome, T2DM, concentrations and 1c), as well as leading to improve- 19 syndrome and involves nonalcoholic fatty liver disease/nonalcoholic steatohepatitis ments in vascular and kidney functions . In patients excessive fat accumulation (NAFLD/NASH) and CVD38. Chronic low-grade with periodontitis who do not have diabetes, periodontal (steatosis) in the liver in the inflammation­ is a unifying feature and contributor to treatment improves endothelial function, as assessed by absence of significant alcohol these disorders38; at least in principle, therefore, per- flow-mediated dilatation42. A longitudinal improvement consumption. NAFLD may progress to NASH, where iodontitis may affect metabolic–inflammatory con­ in periodontal health is linked to attenuated progression of fat accumulation in the liver is ditions by increasing the systemic inflammatory intima–medial thickness of the carotid artery, a marker accompanied by inflammation, burden5,19,22,28–30,32,33. In this regard, periodontitis is of atherosclerosis43. Periodontal treatment of patients leading to fibrosis and associated with elevated serum levels of low-density with periodontitis who are hyperlipidaemic reduces ultimately cirrhosis (advanced 39 scarring/fibrosis) and end-stage lipoprotein (LDL) and triglycerides and reduced levels LDL and triglyceride levels and increases HDL levels . 39 liver failure. of high-density lipoprotein (HDL) . The association of Periodontal treatment of patients with NAFLD/NASH periodontitis, obesity, dyslipidaemia, T2DM, NAFLD/ decreases the serum levels of hepatic–parenchymal HbA 4,39–41 1c NASH and CVD (reviewed in refs ) might be attri­ damage markers, aspartate aminotransferase and HbA , or glycated 1c buted, in part, to causal relationships, as suggested by alanine aminotransaminase44, and in patients with cir- haemoglobin, represents a form of haemoglobin that has clinical interventions. rhosis, periodontal treatment reduces local and systemic glucose attached to it via a non-enzymatic reaction called glycation. The concentration Blood–brain barrier of HbA1c in blood reflects the average levels of blood glucose of the previous 3–4 months.

Therefore, HbA1c is used not only in the diagnosis of Alzheimer disease diabetes but, importantly, as a measure for monitoring glycaemic control in patients Periodontal disease with diabetes. Oro-pharyngeal aspiration Oro-digestive translocation

Endothelial dysfunction Bacteraemias

Cardiovascular disease Aspiration pneumonia Low-grade inflammation Nonalcoholic fatty liver disease Acute phase response

Metabolic alterations Diabetes

Inflammatory bowel disease Colorectal Adverse pregnancy cancer outcomes

Trained myelopoiesis

Priming of osteoclast precursors Periodontal comorbidities

Rheumatoid arthritis Modes of periodontal bacterial dissemination

Host responses linking periodontal disease to comorbidities

Fig. 1 | Periodontal disease and associated inflammatory comorbidities. On the basis of epidemiological, clinical intervention and animal model-based studies, periodontitis has been linked with a number of comorbid conditions, such as those indicated. Mechanistically, periodontitis is associated with bacteraemias and systemic inflammation, which can induce acute-phase responses as well as metabolic and inflammatory alterations in the liver and bone marrow, activities that can influence comorbid conditions. Moreover, periodontal bacteria can disseminate by different routes — haematogenous, oro-pharyngeal and oro-digestive — to reach extra-oral sites where they can cause or exacerbate inflammatory pathologies.

428 | July 2021 | volume 21 www.nature.com/nri Reviews

46 Box 2 | Trained immunity and maladaptive inflammation levels of CRP, IL-6 and LDL cholesterol . Vascular inflam­mation can likely be attributed, in part, to Immunological memory is the ability to recall past pathogen encounters and periodontitis-induced activation of circulating mono- initiate a faster and stronger immune response upon re-challenge; this crucial cytes and their enhanced adhesion to aortic endothelial feature of immunity has been hitherto considered an exclusive prerogative of 17,155 cells via nuclear factor-κB (NF-κB) p65 nuclear trans- adaptive immunity . However, recent studies have shown that earlier microbial 50 or inflammatory challenges (for example, those triggered by damage-associated location and upregulation of VCAM1 in the latter . molecular pattern (DAMP) molecules) imprint a form of memory also in innate immune LIP in rabbits fed a high-cholesterol diet causes forma- cells that enables them, upon returning to a non-activated state, to respond more tion of more aortic atheromatous plaque than occurs in robustly to secondary heterologous challenges. This state of enhanced immune perio­dontally healthy controls under the same diet, and responsiveness is termed trained immunity and is shaped by long-term epigenetic and local treatment of periodontitis with a pro-resolving metabolic alterations, as opposed to permanent genetic changes (gene recombination agent (resolvin E1) attenuates LIP-induced vascular and somatic hypermutation) that characterize classical adaptive immunological inflammation and atherogenesis51. 17,155 memory (see the figure) . Innate immune memory lacks the specificity of its In mice fed a high-fat diet (HFD), oral gavage with adaptive counterpart and can confer broad-based protection against future infections a combination of three human periodontal pathogens with the original or even unrelated pathogens. A trained phenotype (typified by (, Fusobacterium nuclea- epigenetic and transcriptional alterations and heightened immune preparedness) was initially documented in mature myeloid cells; however, the long-lasting (several tum and intermedia) not only induced local months) consequences of trained immunity could not be reconciled with the relatively inflammatory bone loss but also appeared to aggravate short lifespan (a few days) of circulating myeloid cells. This paradox was resolved HFD-induced glucose intolerance and insulin resistance when trained immunity was shown to be initiated in bone marrow haematopoietic and increased the ratio of fat to lean tissue mass9. These stem and progenitor cells (HSPCs). Inflammation-induced metabolic, epigenetic and findings were reproduced in HFD-fed mice subjected transcriptomic adaptations in HSPCs instruct them to undergo enhanced proliferation to continuous low-rate infusion of P. gingivalis LPS, and myeloid-biased differentiation, generating trained or hyper-reactive myeloid cell mimicking chronic low-grade systemic inflammation9. 68,156 populations . Although trained immunity is an evolutionarily conserved function Additionally, periodontal pathogens might potentially 17,155 that promotes host survival upon re-infection , it may boost chronic inflammation aggravate insulin resistance through their metabolic in a detrimental manner, as exemplified in cardiometabolic diseases, in which activities. In a model of P. gingivalis-induced perio- increased bone marrow haematopoietic activity drives persistent leukocytosis and contributes to their chronicity157–160. HSPCs that are epigenetically trained (by earlier dontitis in HFD-fed mice, infection with the wild-type infection or injury) to elicit a heightened myelopoietic response may also exacerbate organism, but not with a branched-chain amino acid other inflammatory or autoimmune conditions with myeloid cell involvement. (BCAA) aminotransferase-deficient mutant, resulted in increased serum levels of BCAAs (such as Leu, Ile Secondary Primary and Val) and insulin resistance, as compared with Trained phenotype (unrelated) challenge 52 challenge uninfected HFD-fed mice . This is consistent with the notion that elevated circulating BCAAs contribute to increased risk of T2DM53. Periodontitis-associated Krebs insulin resistance and the resulting hyperglycaemia cycle may also induce formation of advanced glycation end ↑ Metabolic ↑ Chromatin ↑ Gene products (AGEs), leading to enhanced activation of the activity accessibility expression pro-inflammatory receptor for AGE (RAGE), which is

Response magnitude involved in several metabolic disorders, including vascu­ lar inflammation54 and atherosclerosis55, as well as in Time experimental periodontitis in diabetic mice56. In an HFD-induced NAFLD model in mice, intra- inflammation and endotoxaemia, as well as improving venous administration of P. gingivalis (but not of oral cognitive function and quality of life in those with prior streptococcal species) significantly increased body and hepatic encephalopathy22. Consistently, a cohort study of liver weight and elevated alanine aminotransaminase 6,165 individuals, without baseline liver disease, correlated and liver triglyceride concentrations, compared with periodontitis with incident liver disease independently HFD controls44. Similarly, P. gingivalis infection of mice of confounders45. subjected to NASH appeared to promote steatosis and The notion that periodontitis causes systemic met- hepatic stellate cell activation, thereby causing more abolic complications is supported by studies in rodents severe fibrosis (relative to uninfected NASH controls) or lagomorphs subjected to procedures used as experi- associated with altered fatty acid metabolism in the mental models of periodontitis, such as ligature-induced liver57,58. LIP in rats resulted in steatosis, likely attributed periodontitis (LIP) or oral gavage with human perio- to increased oxidative stress and lipid peroxidation, as dontal pathogens (Box 1). A few examples are given here. suggested by lower glutathione and higher malondi- Similarly to what is seen in human periodontitis, experi­ aldehyde concentrations than those of the unligated mental animal periodontitis causes increased serum controls59. Atheromatous plaque levels of acute phase proteins (such as CRP and Although systemic inflammation induced by blood- Fat accumulation in the intima serum amy­loid A) and inflammatory cytokines (for borne periodontal pathogens may explain, to a signi­ (inner lining) of arteries that example, IL-1β and IL-6)6,46–49 and is thus an appropriate ficant extent, the above-discussed metabolic alterations, results in fibrous thickening model to study the periodontitis connection with sys- direct action of translocated pathogens in the liver can- or calcification of the arteries, restriction of blood flow and temic comorbidities. LIP in rats leads to vascular inflam- not be ruled out. In this regard, in LIP-subjected mice, an enhanced risk of thrombotic mation and endothelial dysfunction that is accompanied viable oral bacteria could be recovered from the liver occlusion. by increased neutrophil numbers and increased serum and spleen until the periodontitis-affected teeth were

NATuRe RevIewS | IMMunOlOgy volume 21 | July 2021 | 429 Reviews

60 Emergency myelopoiesis extracted . Moreover, periodontitis-associated systemic alterations to the gut microbiota and aggravated glucose 61 A tightly regulated process inflammation could also be mediated by swallowed intolerance, insulin resistance and liver steatosis . initiated in response to oral bacteria, which, under certain conditions, cause systemic infection for gut dysbiosis and increased gut permeability with sub- Alterations in the bone marrow de novo generation of 6,36 mature myeloid cells, sequent endotoxaemia and systemic inflammation . The bone marrow is the primary site of haematopoiesis resulting from accelerated In this regard, most of the bacterial species of the gut where all mature blood cell types arise from a common proliferation and enhanced microbiome associated with liver cirrhosis are of oral ancestor, the haematopoietic stem cell (HSC)62. To opti- myeloid differentiation origin35. These important clinical findings suggest bac- mally perform its function, the bone marrow needs to of progenitors in the bone terial translocation from the mouth to the gut, possibly respond swiftly to peripheral challenges such as infection marrow. The goal is to emergency myelopoiesis meet the increased demand facilitated by defective secretion of gastric acid and bile or injury. For instance, in , HSCs 4,35 for neutrophils and other salts in patients with cirrhosis . In the context of the respond rapidly to systemic infection with increased phagocytes that are consumed maladaptation of the oral–gut–liver axis, oral infec- proliferation and myeloid differentiation leading to in large quantities during tion of HFD-fed mice with the periodontal pathogen enhanced myelopoiesis63. In this regard, systemic infec- systemic infections. Aggregatibacter actinomycetemcomitans resulted in tions or inflammation can be sensed in the bone marrow by haematopoietic stem and progenitor cells (HSPCs) Bacterial translocation via Toll-like receptors (TLRs) or receptors for inflam- and bacteraemia matory cytokines, thereby initiating the process for increased production of cells that fight infections, such Bone as neutrophils and monocytes18. marrow Through further differentiation via macrophage colony-stimulating factor (M-CSF) and receptor acti- vator of NF-κB ligand (RANKL), the monocytic line­ age is the precursor of osteoclasts, which are giant multinucleated bone-resorbing cells64. Although osteo­ + + hi ↑ CD11b CSF1R Ly6C clast precursors (OCPs) are generated in the bone osteoclast precursors marrow, they can circulate in the blood and enter sites of active bone resorption, where they can diffe­ rentiate to osteoclasts. Sustained systemic release of P. gingivalis, via a micro-osmotic pump, to simu­ Circulation late periodontitis-associated bacteraemia in mice, ↑ IL-6 resulted in expansion of an OCP population (defined as CD11b+CSF1R+Ly6Chi) with enhanced osteoclasto­ genic lineage bias (relative to the counterpart popu­ lation from uninfected controls)16. This OCP expansion in the bone marrow was dependent on IL-6, the serum levels of which were increased in P. gingivalis-infected mice16 (Fig. 2). In vivo adoptive transfer and tracking Periodontal disease experiments showed that the P. gingivalis-induced OCPs could populate bone resorption sites and differentiate into mature osteoclasts16. Consistently, compared with T 17 cell H those from healthy controls, peripheral blood mono- IL-17 nuclear cells isolated from the blood of patients with periodontitis are more poised to become osteoclasts, Osteoclast precursor as they undergo enhanced RANKL-induced osteoclast 65 Osteoblast RANKL Osteoclast differentiation into mature osteoclasts . Thus, perio- dontal bacteria-induced systemic IL-6 drives the expan- sion of OCPs that traffic to sites of bone resorption to boost osteoclastogenesis in response to locally produced Rheumatoid arthritis RANKL, suggesting that changes in the bone marrow may link periodontitis to other bone loss disorders, such Fig. 2 | Periodontal bacterial translocation leading to bacteraemia and alterations as rheumatoid arthritis (Fig. 2). in the bone marrow that promote osteoclastogenesis in different sites. Periodontal As alluded to above, HSPCs are more sensitive than bacteria may translocate through the ulcerated epithelium of the periodontal pockets traditionally thought to acute infection or chronic inflam­ into the circulation, causing bacteraemia and systemic inflammation. Blood-borne mation, as they express TLRs (for example, TLR4 for Porphyromonas gingivalis causes an increase in the serum levels of IL-6, which sensing LPS) and receptors for inflammatory cytokines in turn induces the expansion of an osteoclast precursor population (defined as and growth factors (such as IL-1β, IL-6 and M-CSF)62,66. CD11b+CSF1R+Ly6Chi) in the bone marrow. This osteoclast precursor population displays enhanced osteoclastogenic lineage bias and populates various bone resorption Such stimuli induce expansion of HSPCs and their sites, where it can differentiate into mature osteoclasts, in response to locally produced skewed differentiation towards the myeloid line­age. receptor activator of NF-κB ligand (RANKL) (for example, produced by osteoblasts Inflammatory adaptation of HSPCs underlies the persis- 18 (Box 2) stimulated by T helper 17 (TH17) cell-derived IL-17). This concept may be a mechanism tence of trained immunity and is regulated by whereby the bone marrow might link periodontitis to other bone loss disorders, such as long-term cell-intrinsic (metabolic, transcriptional and rheumatoid arthritis. epigenetic) mechanisms. IL-1β plays a significant role

430 | July 2021 | volume 21 www.nature.com/nri Reviews

Comorbidities least in part, for the association between periodontal and cardiovascular inflammation. Whether periodontitis can indeed influence hae- Other tissues/organs matopoietic tissue activity and trained immunity in the Periodontal Cardiovascular bone marrow has yet to be addressed experimentally. disease disease However, certain clinical findings may be consistent with Periodontium this notion. Upon ex vivo activation by LPS or whole Recruitment Other risk factors bacteria, peripheral blood neutrophils or monocytes from patients with periodontitis respond with higher Bone pro-inflammatory cytokine production than the same marrow cell types from healthy individuals71,72. Interestingly, this hyper-reactivity often remains even after success- Circulation ful periodontal therapy; that is, the higher capacity to • Bacteraemias produce cytokines seen in myeloid cells from patients • Systemic Hyper-reactive versus healthy controls persists after therapy for at least inflammation myeloid cells 2 months (the time interval for post-treatment review)71,72. These data resemble findings from individuals treated with agents that are associated with induction of trained Bone marrow Trained immunity, such as the Bacillus Calmette–Guérin (BCG) myeloid cells vaccine73,74, and could thus be interpreted to mean that IL-1β the peripheral blood myeloid cells of patients with per- Epigenetic/metabolic rewiring iodontitis are in a trained state, which enables them to induce a heightened response (compared with cells Granulocytes from healthy controls) to future inflammatory chal- lenges. Given that periodontitis affects haematopoietic HSC MPP GMP tissue activity20, this trained state of peripheral mye- Myeloid bias Monocytes loid cells might be the result of long-term epigenetic immune memory imprinted in inflammation-adapted 17,68 Fig. 3 | Trained myelopoiesis in the bone marrow as a mechanistic basis of HSPCs in the bone marrow . Similarly, a subset of inflammatory comorbidities. Periodontitis-associated bacteraemias and systemic patients with rheumatoid arthritis in clinical remission inflammation, for example, inflammatory cytokines, such as IL-1β, can be sensed in the exhibit increased bone marrow and splenic metabolic bone marrow, leading to induction of long-term metabolic and epigenetic rewiring in activity as well as increased arterial wall inflammation haematopoietic stem cells (HSCs). These inflammation-adapted HSCs proliferate together with a promigratory phenotype of circulating and preferentially undergo myeloid-biased differentiation, leading to expansion of monocytes75. Thus, being in remission does not neces- multipotent progenitors (MPPs) with myeloid potential and granulocyte–macrophage sarily normalize the risk of developing CVD in patients progenitors (GMPs) and ultimately production of ‘trained’ myeloid cell populations. with rheumatoid arthritis. Conversely, cardiometabolic These hyper-reactive neutrophils or monocytes/macrophages can be recruited to the conditions were shown to influence haematopoiesis and periodontium and other sites of infection, inflammation or injury, thereby potentially 76–78 exacerbating periodontitis and promoting inflammatory pathology of comorbid trained immunity . conditions such as cardiovascular disease. These findings are consistent with the concept that an inflammatory disease may affect another disease quanti- tatively by enhancing myelopoiesis in the bone marrow in the induction of trained immunity in humans and or qualitatively by driving generation of myeloid cells mice67,68 (Fig. 3). In the bone marrow, this cytokine acts with enhanced inflammatory potential, even in patients on HSPCs and promotes, likely via changes in their cellu­ in remission (implying the presence of innate immune lar metabolism, a myeloid-differentiation bias, which is a memory in HSPCs). Regardless of whether the bone hallmark of HSPCs in trained immunity18,68,69. marrow is first affected by periodontitis-induced sys- As a cause of systemic inflammation, periodontitis temic inflammation or systemic inflammation caused by might, at least in principle, trigger inflammatory adapta­ another condition, such as CVD, the resulting heightened tion of HSPCs and trained immunity in the bone mar­ and sustained (‘trained’) myelopoiesis can have an impact row. This notion is backed by a recent clinical imaging on both conditions (Fig. 3). The production of trained/ study using 18F-fluorodeoxyglucose positron emission hyper-reactive myeloid progeny and their recruitment to 18 Rheumatoid arthritis tomography–computed tomography ( F-FDG-PET–CT). inflamed tissues with consequent exacerbation of local A chronic inflammatory This study showed a correlation between metabolic pathology could possibly generate a positive-feedback autoimmune disease acti­vity within the periodontium (surrogate marker of loop between inflammation-adapted haematopoietic typified by production of autoantibodies to various perio­dontal inflammation) and haematopoietic tissue progenitor cells in the bone marrow and the inflam- 18 targets (for example, activity (reflecting progenitor cell activity and stimulated matory disorders (Fig. 3). According to this concept, citrullinated matrix proteins) myelopoiesis), as well as with arterial inflammation20. periodontitis can affect comorbid conditions and vice and induction of inflammation An earlier 18F-FDG-PET–CT study indicated a correla­ versa. Consistent with the latter, in a prospective study that causes progressive erosion tion between periodontal inflammation and/or meta­ with 1,784 subjects and an 11‐year follow‐up, serum of cartilage and bone in the joints. Its aetiology is complex bolic activity and histologically confirmed carotid fibrinogen and white blood cell count (markers of sys- 70 and includes both genetic and plaque inflammation . Together, these findings suggest temic inflammation) were longitudinally associated with environmental factors. that increased haemato­poietic activity may account, at periodontitis severity in a dose‐dependent manner79.

NATuRe RevIewS | IMMunOlOgy volume 21 | July 2021 | 431 Reviews

Keystone pathogen Peripheral blood neutrophils from patients with per- induced by systemic inflammation (due to treatment A pathogen that exerts a iodontitis display increased expression of interferon- with purified agonists or perhaps associated with perio- disproportionately large effect stimulated genes that correlates with increased plasma dontitis or other condition) might lead to the genera­tion on its community relative to IFNα concentrations, relative to healthy individuals80. of hyper-reactive neutrophils, which, depending on the its abundance, forming the ‘keystone’ of the community’s Ex vivo, these neutrophils exhibit a hyper-reactive disease context, can be protective (for example, in tumour structure; for instance, in phenotype in terms of reactive oxygen species (ROS) immunity) or destructive (for instance, in inflammatory experimental periodontitis, production when primed with type I interferons (but conditions). low-abundance Porphyromonas not LPS)80. Whether this type I interferon-associated gingivalis modulates the size hyper-reactivity of peripheral blood neutrophils from Direct effects of periodontal bacteria and composition of the local microbial community in a patients with periodontitis arises from trained granulo- Multiple clinical studies have detected genomic DNA manner that promotes poiesis in the bone marrow is uncertain. Interestingly, from periodontal bacteria in systemic tissues; however, dysbiosis. mice systemically treated with fungally derived β-glucan, there is limited evidence for their presence in a viable a known inducer of trained immunity, display type I state82,83. Although periodontal bacteria have evolved interferon-dependent trained granulopoiesis with tran- immune subversive strategies that enable them to thrive scriptomic and epigenetic rewiring of granulopoietic in periodontal pockets3, circulating periodontal path- progenitors in the bone marrow, leading to the genera­ ogens may be more vulnerable to phagocytic killing tion of neutrophils with an enhanced ROS-dependent in the blood or in systemic tissues. Nevertheless, even if tumour-killing phenotype81. Thus, trained granulopoiesis the presence of viable periodontal bacteria in systemic tissues is transient, their effects — for instance, their release of virulence factors, such as toxic proteases11,15, or their induction of inflammation — can be important, given the chronicity of periodontitis and the frequency of bacteraemias, which occur not only during profes- sional dental care (for example, debridement of the T 17 cell H pathogenic dental biofilm, tooth extractions) but also Mouth during routine activities (such as tooth brushing or mastication)21. Periodontal bacteria, such as P. gingivalis, Draining are also detected (by 16S ribosomal DNA sequencing) Oro-digestive lymph node in blood myeloid-derived dendritic cells (mDCs)84. route Bacteraemias induced by debridement elevate the Trafficking to gut P. gingivalis content of blood mDCs and the frequency of the carrier mDCs, which have been shown by immuno­ histochemistry to colocalize with P. gingivalis in athero­ matous plaques of patients with periodontitis. As P. gingivalis may survive within mDCs in vitro, this mechanism might serve to disseminate the pathogen to remote sites, such as atheromas84. Non-haematogenous Intestine Inflammation Antigen dissemination of periodontal bacteria has also been presentation described. Swallowed periodontal bacteria may trans- locate via the oro-digestive route and ectopically colo- nize the gut (Fig. 4); indeed, periodontal bacteria have APC been detected in the gut of patients with IBD8,85,86. IL-1β Interestingly, periodontal bacteria may not neces­ sarily have to relocate from the periodontium to affect Proliferation Macrophage pathology in distant sites, such as the joints2,12. Below, IL-1β we consider in more detail the effects of disseminated periodontal bacteria as well as the effects of locally acting periodontal bacteria on distal tissues (Table 1).

Endothelial dysfunction. The gingipain proteases of the TH17 cells periodontal keystone pathogen P. gingivalis are critical for Fig. 4 | Oral–gut axis mechanisms that promote colitis. Periodontitis-associated its capacity to colonize subgingival tooth sites, procure pathobionts can reach the gut through the oro-digestive route (owing to constant saliva nutrients, subvert immunity and contribute to perio- swallowing) and can promote colitis in susceptible hosts, in part through induction of dontal tissue destruction2,87. Gingipains are important IL-1β by inflammatory macrophages. Oral pathobiont-reactive T cells (enriched in T helper also for the virulence of P. gingivalis in extra-oral sites. 17 (T 17) cells), which expand during periodontitis, migrate through the lymphatics to H In a zebrafish larva infection model, P. gingivalis induced the gut, where they are activated by the ectopically colonized oral pathobionts upon their processing by antigen-presenting cells (APCs). The oral pathobiont-induced oedema by increasing vascular permeability, which was elevation of IL-1β contributes to the activation and proliferation of the transmigrated attributed to gingipain-dependent degradation of plate- let endothelial cell adhesion molecule (PECAM1) and TH17 cells, which become colitogenic and exacerbate intestinal inflammation. Thus, ectopically colonized oral pathobionts may exacerbate colitis by activating both innate vascular endothelial cadherin (VE-cadherin) disrupting immunity (local inflammatory macrophages) and adaptive immunity (transmigrated the endothelial barrier15 (Fig. 5a). Consistently, human

TH17 cells of oral origin). endothelial cells infected by P. gingivalis (but not by

432 | July 2021 | volume 21 www.nature.com/nri Reviews

Table 1 | Potential mechanisms of human periodontal bacterial involvement in extra-oral pathologies Bacterium Effector Target Outcome Refs molecule Porphyromonas gingivalis Gingipains PECAM1, VE-cadherin, in vascular Impairment of endothelial junctional integrity, vascular 15 endothelial cells damage Gingipains Tau protein, APOE in the brain Alterations promoting neurotoxic/inflammatory effects 11 PPAD Proteins, including host molecules, Citrullination of protein autoantigens leading to ACPA 125–127 such as fibrinogen and α-enolase induction Mfa1 fimbrial DC-SIGN on DCs Hijacking the trafficking of DCs; enhanced atherogenic 13,84 protein and anti-apoptotic potential BCAT Respective α-keto acidsa Increased plasma levels of BCAAs (Leu, Ile and Val); 52 insulin resistance Aggregatibacter Leukotoxin A β2 Integrin on neutrophils PAD hyperactivation and release of citrullinated 12 actinomycetemcomitans autoantigens through cytolysis, leading to ACPA induction 8 Klebsiella pneumoniae Unspecified TLRs on colonic epithelial cells and DCs IL-18 secretion, TH1 cell expansion and gut inflammation TLR agonists Fusobacterium Fap2 Gal-GalNac on CRC cells CRC targeting and enrichment; induction of 7,113 nucleatum pro-metastatic chemokines Fap2 TIGIT on T cells and NK cells Tumour immune evasion 10 FadA E-cadherin on CRC cells Activation of Wnt–β-catenin signalling and CRC cell 112,118 proliferation ACPA, anti-citrullinated protein antibody; APOE, apolipoprotein E; BCAA, branched-chain amino acid; BCAT, branched-chain amino acid aminotransferase; CRC, colorectal cancer; DC, dendritic cell; DC-SIGN, DC-specific ICAM3-grabbing non-integrin; NK, natural killer; PAD, peptidyl-arginine deiminase;

PECAM1, platelet endothelial cell adhesion molecule; PPAD, Porphyromonas gingivalis peptidyl-arginine deiminase; TH1 cell, T helper 1 cell; TIGIT, T cell immunoreceptor with immunoglobulin and ITIM domains; TLR, Toll-like receptor; VE-cadherin, vascular endothelial cadherin. aα-ketoisocaproate (Leu), α-keto-β-methylvalerate (Ile) and α-ketoisovalerate (Val).

a gingipain-deficient mutant) displayed decreased cell and accumulation of insoluble tau forms in Alzheimer surface expression of PECAM1 and VE-cadherin, dis- disease92 (Fig. 5b). rupted cell–cell junctions and elevated endothelial cell Consistent with these clinical and laboratory obser- permeability, which was reversed by a specific gingi- vations, the brains of aged (11-month-old) mice, which pain inhibitor15. Vascular damage and compromised were orally infected with P. gingivalis, showed DNA- endothelial junctional integrity may potentially trigger based evidence of pathogen infiltration and increased 11 platelet aggregation, leukocyte adhesion and induction levels of amyloid-β1–42, a component of amyloid plaques (Fig. 5b) of pro-inflammatory cytokines, thereby possibly pro- . By contrast, amyloid-β1–42 was not elevated moting atherosclerosis88 (Fig. 5a). It is uncertain whether (relative to the baseline of mock infection) when mice P. gingivalis-induced vascular damage also occurs in were infected with gingipain-deficient P. gingivalis humans and contributes to atherosclerotic lesions in sus- or with wild-type P. gingivalis together with an orally ceptible individuals; intriguingly, however, P. gingivalis is the administered gingipain inhibitor11. Gingipain inhibition most abundant bacterial species detected in non-diseased also reduced the P. gingivalis load and neuroinflam- vascular tissue of patients with atherosclerosis89. mation and prevented loss of hippocampal neurons11. Gingipains It remains uncertain how P. gingivalis gains access to the A family of Porphyromonas Alzheimer disease. Several epidemiological studies brain, although its ability to cause increased vascular gingivalis-derived trypsin-like have suggested an association between periodonti- endothelial permeability15 (Fig. 5a) may facilitate its cysteine proteases that make Alzheimer disease ref.90 a major contribution to its tis and (reviewed in ). A recent passage through the blood–brain barrier. virulence. High-molecular large retrospective cohort study associated clinical and The apolipoprotein E gene (APOE) (particularly the mass arginine-specific gingipain microbial markers of periodontitis with Alzheimer E4 allele) is the strongest genetic risk factor for late-onset A (HRgpA), arginine-specific disease incidence and mortality91. A plausible mech- Alzheimer disease. APOE is implicated in the metab­olism, gingipain B (RgpB) and anism underlying the link between periodontitis and aggregation and deposition of amyloid-β93. APOE frag- lysine-specific gingipain 94 (Kgp) are the members Alzheimer disease might be the infiltration of the brain mentation may have neurotoxic effects , and P. gingivalis 95 of the gingipain family. with periodontal pathogens, a hypothesis supported by gingipains cleave APOE at Arg residues , suggesting emerging evidence implicating P. gingivalis in particu- an additional possible mechanism whereby P. gingivalis Alzheimer disease lar. P. gingivalis DNA has been detected in Alzheimer might contribute to Alzheimer disease. P. gingivalis can Progressive neuroinflammatory and neurodegenerative brain disease brain autopsy specimens and the cerebrospinal cause neuroinflammation also in the absence of APOE. −/− disease, characterized by fluid of living individuals with likely Alzheimer disease Oral infection of ApoE mice with P. gingivalis, but extraneuronal deposition of diagnosis11. Also detectable in Alzheimer disease brain not with two other important periodontal pathogens amyloid-β peptides (neuritic autopsies are the pathogen’s gingipains, the abundance ( and ), resulted plaques) and intraneuronal of which correlates with tau protein load and ubiqui- in brain infection, complement activation and neuronal accumulation of hyperphos- 11 96 (Fig. 5b) phorylated and fragmented tin load, markers of Alzheimer disease pathology . injury . Aberrant complement activation, which 11 97 tau, a microtubule-associated In vitro, the gingipains can cleave tau , an activity that is associated with Alzheimer disease pathogenesis , may protein (neurofibrillary tangles). might contribute to aberrant phosphorylation of tau be facilitated in the absence of APOE (or in the presence

NATuRe RevIewS | IMMunOlOgy volume 21 | July 2021 | 433 Reviews

a Circulation Cytokine Platelet secretion aggregation Intact Endothelial cells junction

Increased permeability

Gingipains

• Plasma leakage VE-cadherin VE-cadherin • Enhanced leukocyte extravasation PECAM1 PECAM1

b Damaged neuron Complement activation

Neurofibrillary ↑ Amyloid-β Gingipain (tau) tangle plaque Neuronal Tau inflammation injury

Amyloid-β plaque Tangles of fragmented/ hyperphosphorylated Tau

c P. gingivalis A. actinomycetemcomitans

LtxA Gingipains Ca2+ influx Cytolysis PPAD

Host PAD proteins hyper- Neutrophil activation NETosis Periodontium

Citrullinated Citrullinated epitope host proteins

APCs, T cells, HLA-DRB1 SE B cells, plasma cells

ACPA

of defective APOE). Indeed, human APOE binds with titres at baseline involves clusters of bacteria rather than high affinity to activated C1q, the initiator of the classi- individual bacteria91. In addition to P. gingivalis, such cal complement cascade, and the resulting C1q–APOE clusters include Campylobacter rectus and Prevotella complexes (which are detected in amyloid-β plaques) melaninogenica among other bacteria91. Thus, an downregulate complement activity98. interesting question is whether and, if so, how P. g in- The risk of Alzheimer disease-related mortality in givalis cooperates with other bacteria to possibly con- individuals 65 years or older is increased when the mor- tribute to Alzheimer disease pathology, as it does in tality association with anti-periodontal pathogen IgG periodontitis87.

434 | July 2021 | volume 21 www.nature.com/nri Reviews

◀ Fig. 5 | Mechanisms of periodontal bacterial action in extra-oral pathologies. pathobionts (predominantly comprising Klebsiella spp. a | Porphyromonas gingivalis causes vascular endothelial barrier disruption and increased and Enterobacter spp.) was shown to ectopically colonize the permeability by disrupting intercellular junctions. This action is mediated by its gingipain mouse gut, where they promoted colitis through induction proteases, which degrade platelet endothelial cell adhesion molecule (PECAM1) and of non-canonical inflammasome-dependent IL-1β secretion vascular endothelial cadherin (VE-cadherin), which are crucial for junctional integrity. lo hi hi Endothelial damage and elevated permeability may instigate several processes that by inflammatory macrophages (Ly6C MHC-II CD64 ) potentially induce or exacerbate atherogenesis, including induction of pro-inflammatory (Fig. 4). Gut colonization by oral pathobionts and promo­ cytokines, platelet aggregation and increased leukocyte extravasation to subendothelial tion of colitis required pre-existing intestinal inflam­ areas. b | P. gingivalis-induced pathology in Alzheimer disease. P. gingivalis DNA and mation (induced by dextran sodium sulfate treatment) gingipains have been detected in brain autopsies of patients with Alzheimer disease as or a colitis-susceptible host, specifically IL-10-deficient well as in the brain of mice orally infected with this pathogen. The presence of P. gingivalis mice6, which mimic IBD-susceptible humans with specific in the mouse brain is associated with increased levels of amyloid-β, complement IL-10R polymorphisms103. activation and neuroinflammation. The gingipains of the pathogen cleave the microtubule- Human oral pathobionts can also promote colitis associated protein tau, an activity that promotes aberrant phosphorylation of tau and in animal models. Germ-free mice gavaged with oral accumulation of misfolded insoluble tau in Alzheimer disease. c | Role of periodontal (salivary) microbiota from a patient with Crohn’s dis- bacteria in the generation of anti-citrullinated protein antibodies (ACPAs). P. gingivalis expresses a unique (among prokaryotic organisms) peptidyl-arginine deiminase (PPAD), ease displayed expansion of intestinal T helper 1 (TH1) + + which can citrullinate proteins including host proteins. The PPAD activity is facilitated IFNγ CD4 cells, which was attributed to Klebsiella spp. 8 by the pathogen’s arginine-specific gingipains, which cleave proteins and expose (Klebsiella pneumoniae and Klebsiella aeromobilis) . C-terminal arginine residues for citrullination by PPAD. Alternatively, Aggregatibacter Ectopic gut colonization by oral K. pneumoniae 2H7

actinomycetemcomitans indirectly causes host protein citrullination by secreting invariably resulted in expansion of TH1 cells, although leukotoxin A (LtxA), a pore-forming toxin that induces calcium influx and hyperactivation induction of gut inflammation additionally required of PAD enzymes in neutrophils, as well as cytolysis (by NETosis, a form of cell death a colitis-susceptible host, specifically germ-free typified by the release of decondensed chromatin and granular contents to the IL-10-deficient mice8. The K. pneumoniae 2H7-induced extracellular milieu), thereby releasing the generated citrullinated autoantigens. Thus, T 1 cell response required the presence of the intes­ through distinct mechanisms, both pathogens can contribute to the generation of the H tinal CD11b–CD103+ DC subset, possibly acting as rheumatoid arthritis-specific ACPAs that promote disease in individuals with HLA-DRB1 shared epitope (SE) alleles. APC, antigen-presenting cell. antigen-presenting cells, and TLR4-activated colonic epithelial cells secreting IL-18 appeared to amplify the 8 TH1 cell response . It should be noted that P. gingivalis may not be an As discussed above, LIP in wild-type mice does obligatory mechanism for the connection between per- not lead to gut colonization by oral pathobionts in the iodontitis and Alzheimer disease. In a mouse model of absence of pre-existing colitis pathology6. Moreover, Alzheimer disease, induction of experimental perio- wild-type mice are resistant to colonization by K. pneu- dontitis in the absence of P. gingivalis (by using the LIP moniae 2H7 unless they are treated with antibiotics8. model; Box 1) increased the levels of insoluble amyloid-β These findings suggest that a normal gut flora can pre- Pathobionts and affected neuroinflammation, both of which vent ectopic colonization by oral bacteria, unless it is dis- Organisms that are generally were attenuated by anti-inflammatory treatment99,100. rupted by local inflammation or by antibiotics. Similarly, benign or commensal but that Although direct involvement of indigenous mouse oral although humans daily swallow 1.5 l of saliva contain- have the capacity to promote 12 36,104 pathology under specific bacteria cannot be formally ruled out (in a manner ing approximately ~10 oral bacteria , oral micro- conditions of disrupted similar to that of P. gingivalis), these findings raise the organisms are generally poor colonizers of the healthy host–microbiota homeostasis non-mutually exclusive possibility that periodontitis human intestine; by contrast, oral microorganisms are (for example, resulting from could contribute to Alzheimer disease pathology also enriched in the gut microbiota of patients with IBD, immune deficiencies, antibiotic by increasing systemic inflammation. colon cancer, liver cirrhosis and other diseases35,85,86,104,105. treatment, tissue damage or dietary shifts). In fact, most of the dominant species with increased Inflammatory bowel disease. IBD is associated with abundance in paediatric Crohn’s disease or ulcerative Non-canonical increased risk of periodontitis101. Consistently, SAMP1/ colitis are normally inhabitants of the oral cavity (such inflammasome YitFc mice, a spontaneous model of Crohn’s disease, as Fusobacteriaceae and Veillonellaceae) rather than Inflammasome is a cytosolic, 102 85,86 multiprotein complex that develop natural periodontitis . Recent animal studies typical intestinal organisms . Importantly, increased activates caspases suggest that periodontal pathogens can contribute to abundance of oral bacteria correlates with disease seve­ (predominantly caspase 1) in intestinal inflammation by influencing gut dysbio- rity, suggesting a potential role of oral species in IBD response to infection or injury. sis and barrier function6,8,9,36, implying a bidirectional pathogenesis or progression85,86. This in turn results in cleavage association between the two disorders. and secretion of inflammatory cytokines, such as IL-1β and Mouse periodontitis induced by oral gavage with P. g in - Colorectal cancer. The gut microbiota of patients with IL-18. Non-canonical is a givalis alone or with other periodontal pathogens (F. nucle- colorectal cancer (CRC) exhibits increased species caspase 1-independent but atum and P. intermedia) alters the gut microbiota9,36 and richness and abundance (compared with healthy con- caspase 11-mediated pathway results in reduced expression of epithelial tight-junction trols), and this is attributed in part to ectopic coloniza- of inflammasome activation, which is crucial for controlling proteins in the ileum and increased expression of pro- tion and expansion of bacteria deriving from the oral 36 Gram-negative bacterial inflammatory cytokines in the colon . As P. gingivalis cavity (for instance, F. nucleatum, Parvimonas micra, infections and can also trigger was only transiently detected in the gut, where it failed Peptostreptococcus stomatis, Solobacterium moorei and caspase 1 activation and to colonize36, the precise underlying mechanisms in this Porphyromonas asaccharolytica)105–108. The evidence subsequent maturation and pioneering study are largely uncertain. for involvement in human CRC is particularly strong for secretion of IL-1β and IL-18. 109,110 Both pathways may induce a Advanced mechanistic understanding was achieved F. nucleatum , an oral bacterium that has also been specific form of cell death, in a recent study that used the LIP model, which leads to implicated in pregnancy complications in both humans called pyroptosis. expansion of indigenous oral pathobionts6. A subset of oral and experimental mouse models, where it can reach the

NATuRe RevIewS | IMMunOlOgy volume 21 | July 2021 | 435 Reviews

fetal–placental compartment by the haematogenous and α-enolase exposing C-terminal arginine residues route111 (Fig. 1). Specifically, F. nucleatum can promote that can be citrullinated by P. gingivalis PAD (PPAD)127. the growth112 and migration113 of human CRC cells and In patients with ACPA-positive rheumatoid arthritis, their resistance to chemotherapy114, as well as enhance the detection of P. gingivalis in the periodontal biofilm intestinal tumorigenesis in the ApcMin/+ mouse model107. correlates with increased levels of ACPAs121. Infection The notion that CRC-associated F. nucleatum origi­ of mice with wild-type, but not PPAD-deficient, nates from the oral cavity is supported by findings strains of P. gingivalis aggravates collagen- or collagen that identical strains of this organism are detected antibody-induced arthritis, accompanied by the pres- in the saliva and colorectal tumours of patients with ence of citrullinated proteins at the infection site and CRC115. Oral F. nucleatum can haematogenously reach generation of ACPAs125,126. Antibiotic treatment of mice the colon116, where it exhibits tropism for CRC. This is subjected to P. gingivalis-induced periodontitis inhibited because CRC cells overexpress a carbohydrate moiety subsequent induction of collagen-induced arthritis in the (Gal-GalNAc) that is bound by the F. nucleatum lec- same animals128. However, it has also been proposed that tin Fap2 (ref.7), which also induces the release of pro- orally administered P. gingivalis can aggravate experi­ metastatic chemokines (IL-8 and CXC-chemokine mental rheumatoid arthritis in mice independently of ligand 1 (CXCL1)) that drive CRC migration113 and its citrullinating activity, presumably by altering the gut tumour-related angiogenesis117. Another F. nucleatum microbiota, impairing gut barrier function and elevating adhesin, FadA, stimulates the growth of CRC (but not of systemic inflammation14,129. non-cancerous) cells by inducing E-cadherin-mediated Although A. actinomycetemcomitans does not encode Wnt–β-catenin signalling in a manner dependent on a PAD-like enzyme, it induces hypercitrullination of host annexin A1, which is upregulated in CRC112,118. Moreover, proteins indirectly by secreting a pore-forming toxin, F. nucleatum alters the tumour microenvironment in a leukotoxin A (LtxA), that binds to β2 integrin adhe- manner that facilitates CRC progression; specifically, sion receptors on neutrophils inducing calcium influx it recruits tumour-infiltrating myeloid-derived suppres- and aberrant hyperactivation of the neutrophil’s PAD sor cells107 and undermines the cytotoxic function of enzymes12. Moreover, LtxA causes the release of citrulli- tumour-infiltrating T cells and natural killer cells10. In the nated autoantigens through cytolysis12 (Fig. 5c). Because latter mechanism, F. nucleatum uses its Fap2 adhesin to the toxicity of LtxA is restricted to leukocytes from pri- activate the inhibitory immunoreceptor T cell immuno­ mates, it has not been possible to test whether A. actino- receptor with immunoglobulin and ITIM domains mycetemcomitans can induce an ACPA response in vivo (TIGIT), thereby enabling the tumour cells to evade killing using currently available rodent models. However, the by T cells and natural killer cells10. Besides F. nucleatum, human relevance of the proposed mechanism is sup- several other oral bacteria are suspected to play a role in ported by findings that the concentration of anti-LtxA different types of carcinoma (reviewed in ref.119). antibodies is strongly associated with ACPA positivity, especially when the patients are analysed in the context Locally acting periodontal bacteria with remote conse- of HLA-DRB1 shared epitope alleles12. quences. Periodontitis is associated with increased dis- ease activity in patients with rheumatoid arthritis120,121, Trafficking of orally primed inflammatory cells and local periodontal treatment reduces the severity Lymphocyte trafficking. The concept that immune of arthritis122–124. Although periodontitis-associated responses primed at a certain mucosal barrier site can systemic inflammation likely contributes to this asso- be effectively recalled at other mucosal sites is a hallmark ciation, two periodontal pathogens may independently of the common mucosal immune system and forms the contribute to the periodontitis–rheumatoid arthritis basis of mucosal vaccination-induced protection against connection. Specifically, through distinct mechanisms, pathogens at both local and remote sites130,131. However, P. gingivalis and A. actinomycetemcomitans may pro- the same mechanism may have detrimental effects and mote the generation of the anti-citrullinated protein link two distinct mucosal inflammatory diseases. antibodies (ACPAs) that are seen in patients with Indeed, T cell priming in the periodontal tissue rheumatoid arthritis2,12,125,126 (Fig. 5c). The disruption of and subsequent T cell trafficking to the gut links per- immune tolerance to citrullinated proteins in rheuma- iodontitis to colitis in mice6 (Fig. 4). Specifically, oral toid arthritis-susceptible individuals is associated with pathobiont-specific T cells, which expand during LIP, HLA-DRB1 alleles encoding a shared epitope that binds express the gut-homing molecules CC-chemokine recep- selectively to citrullinated peptides. Citrullinated pro- tor 9 (CCR9) and α4β7 integrin (the receptor for the teins are found in higher abundance in the inflamed gut-specific vascular addressin MAdCAM1) and migrate periodontium than in healthy tissue, and ACPAs are from the oral cavity-draining cervical lymph nodes to detectable in the circulation prior to rheumatoid arthritis the gut6. Once in the gut, these oral pathobiont-specific Citrullination 2,12 Post-translational modification symptom onset and correlate with disease severity . T cells, in which TH17 cells are enriched, proliferate and 6 of a protein involving P. gingivalis is uniquely equipped, among tested exacerbate colitis (Fig. 4). Interestingly, a monoclonal deimination of arginine prokaryotes, with a peptidyl-arginine deiminase (PAD) antibody to the α4β7 integrin (vedolizumab), which has by peptidylarginine that citrullinates proteins including human fibrinogen been approved for the treatment of IBD132, might in part deiminase to generate and α-enolase, which are major rheumatoid arthri- protect against IBD by inhibiting the trafficking of orally citrulline. Uncontrolled 2,127 citrullination may result tis autoantigens in their citrullinated form . The primed T cells to the gut. in neoepitopes that induce P. gingivalis citrullination capacity is facilitated by its Once in the gut, these T cells of oral origin prolifer- autoantibody generation. arginine-specific gingipains, which cleave fibrinogen ate in response to their cognate antigens, because — as

436 | July 2021 | volume 21 www.nature.com/nri Reviews

discussed above — oral pathobionts can ectopically col- origin, might contribute to the pathological repro- onize the gut6. Moreover, the ectopic oral pathobionts gramming of these cells. However, as the expansion stimulate intestinal macrophages to secrete IL-1β, which of these carrier mDCs in human blood increases sig- contributes to the expansion of the transmigrated oral nificantly after debridement-induced bacteraemia84, T cells6 (Fig. 4). In the gut, these colitogenic T cells of periodontitis-associated pathogens are likely important + + + oral origin display RORγt and RORγt T-bet TH17 cell for their altered pathogenic potential. phenotypes and, therefore, secrete both IL-17A and IFNγ6. This experimental finding is consistent with Conclusions and outlook clinical observations showing the presence of IL-17A Epidemiological, clinical interventional and exper-

and IFNγ and both TH17 cells and cells with a mixed imental studies collectively offer sufficient evidence

TH17/TH1 phenotype in the intestinal mucosa of patients that periodontitis adversely impacts systemic health with IBD133. through biologically plausible mechanisms (Figs 2–5). Consistent with the study discussed above6 and the Although clinical intervention studies suggest that local concept that the mucosal immune system is an inte- periodontal treatment decreases the serum levels of grated, system-wide organ131, an earlier study showed inflammatory factors and improves metabolic control, that respiratory influenza infection in mice results lipid profile and other surrogate markers of systemic in intestinal immune injury attributed to the recruit- disease5,19,22,29,30,32,33,42,43,122,123, clear evidence that successful ment of CCR9+CD4+ T cells from the lungs134. The treatment of periodontal disease can reduce the risk or lung-derived transmigrated T cells caused intestinal incidence of epidemiologically associated conditions is dysbiosis and inflammation by secreting IL-17A and lacking. Such evidence would require multicentre random­ IFNγ134. Although not all individuals with flu, or other ized controlled trials and could establish that perio­ respiratory infections, develop diarrhoea or other gastro- dontitis is a modifiable risk factor for life-threatening enteritis symptoms, this mechanism might contribute to comorbidities. In this regard, further improvement of those substantial cases presenting with both respiratory local periodontal therapy through adjunctive host- and gastroenteritis symptoms, as seen in many patients modulation treatments (such as by targeting complement with COVID-19 (ref.135). or pro-resolution pathways140) may contribute to reduc- ing systemic inflammation and promote systemic health. Myeloid cell trafficking. As alluded to above, blood-borne Moreover, targeted therapies against specific perio­ P. gingivalis is taken up by human blood CD1c+CD209+ dontal pathogens (such as P. gingivalis and F. nucleatum) mDCs through an interaction between its Mfa1 fim- or the key virulence factors they produce (for exam- brial protein and the C-type lectin DC-specific ICAM3- ple, gingipains and Fap2) might ameliorate specific grabbing non-integrin (DC-SIGN)84. Interestingly, comorbidities, such as Alzheimer disease and CRC. P. gingivalis appears to modulate the trafficking and Future translational research should importantly behaviour of mDCs in a manner consistent with the target the mechanistic understanding of the connection detection of P. gingivalis-carrying mDCs in atheromatous between periodontitis and comorbidities. The notion plaques84. In response to internalized P. gingivalis, human that inflammation-adapted haematopoietic progeni- monocyte-derived DCs selectively upregulate the expres- tors may link periodontitis and CVD is consistent with sion of CXC-chemokine receptor 4 (CXCR4), which clinical-imaging correlation of periodontal inflamma- directs homing to atheromatous plaques, whereas they tion with haematopoietic tissue activity and arterial do not upregulate the CCR7, which mediates homing to inflammation20 but will require experimental confirma- secondary lymphoid organs136,137. Consistently, periph- tion in a broader context involving additional comor- eral blood mDCs from patients with periodontitis display bidities (for example, rheumatoid arthritis, T2DM and higher CXCR4 and lower CCR7 expression relative to NAFLD). Blocking key mediators of this process may their counterparts from healthy controls136. Moreover, reduce the risk of multiple comorbidities. Given that human monocyte-derived DCs that have internalized IL-1β is crucial for the induction of innate immune Clonal haematopoiesis of indeterminate potential P. gingivalis acquire enhanced pro-inflammatory and memory, the generation of inflammation-adapted (CHIP). Age-related acquisition athero­genic potential (for example, they secrete high HSPCs and enhanced myelopoiesis in cardiometa- of somatic mutations, which levels of MMP9, an indicator of plaque rupture risk)84. bolic disease18, systemic inhibition of this cytokine confer haematopoietic The P. gingivalis-carrier CD1c+CD209+ mDCs more- may protect not only against atherosclerotic CVD (as stem cell clonal expansion over downregulate apoptosis13, expand in the blood of shown by the CANTOS trial141) but also against sev- advantage and are associated 84 with increased risk of myeloid patients with periodontitis and promote systemic TH17 eral inflammatory comorbidities. Inhibition of IL-1β 138 malignancies, cardiovascular cell responses , thereby potentially influencing not only may potentially disrupt a common mechanistic basis disease and all-cause CVD but also IL-17-driven systemic conditions. The for inflammatory comorbidities, namely the maladap- mortality. The generated inhibition of DC apoptosis (which is generally linked tive training of bone marrow progenitors. Intriguingly, mutant myeloid cells 139 comprise a disproportionately to aberrant inflammatory conditions ) by P. gingivalis IL-1β and IL-6 also seem to link the age-related large fraction of leukocytes depends on MFa1 fimbria-induced activation of a clonal haematopoiesis of indeterminate potential (CHIP) (in peripheral blood and DC-SIGN–AKT–mTOR anti-apoptotic pathway13. to exacerbation of inflammation in CVD142,143 and per- tissues) and display a hyper- The microbial cargo of P. gingivalis-carrier blood haps other ageing-related inflammatory disorders, such inflammatory phenotype, mDCs contains additional species (Burkholderia cepacia, as periodontitis18. CHIP is likely both a driver and a suggesting that CHIP may 18,142,143 also be related to chronic Helicobacter pylori, Pseudomonas spp., Moraxella consequence of inflammation and, as such, may 84 inflammatory diseases catarrhalis, K. pneumoniae and Salmonella enterica) . be associated with periodontitis, although this has yet in general. Therefore, additional species, not necessarily of oral to be addressed by clinical and experimental studies.

NATuRe RevIewS | IMMunOlOgy volume 21 | July 2021 | 437 Reviews

At least in principle, reversing the pathological effects of immunity and CHIP), future clinical studies investi- CHIP may ameliorate the severity of periodontitis and gating the effects of improving periodontal health on associated systemic comorbidities. systemic comorbidities should identify and use surro- The emerging findings that inflammatory processes gate markers of those mechanisms (for instance, mye- in various peripheral tissues (‘peripheral’ inflammation) lopoiesis output in the peripheral blood) and novel may be interlinked via inflammatory adaptations of the risk-assessment scores. We clearly need unifying frame- bone marrow as a central hub (‘central’ inflammation)18 works, in which peripheral inflammatory and infectious may not only improve our mechanistic understanding of stimuli (such as those associated with periodontitis) and comorbidities but also revolutionize the way scientists multiple systemic risk factors (such as, obesity, dyslipi- view inflammation and disease. Specifically, factors regu- daemia, glycaemic level and ageing) can be productively lating peripheral inflammation may in fact do so by addi- integrated to better understand the interconnected tionally modulating central inflammatory processes. For pathogenesis of distinct chronic inflammatory diseases instance, the homeostatic protein developmental endothe- and/or malignant disorders. Inflammatory adaptation of lial locus 1 (DEL1; also known as EGF-like repeat and haematopoietic progenitors represents such a unifying discoidin I-like domain-containing protein 3), which was framework between peripheral and central inflammation shown to inhibit initiation of periodontal inflammation that could also provide the platform for novel therapeu- and to promote its resolution144–147, may also regulate cen- tic interventions targeting inflammation and inflamma- tral inflammatory processes via its function to modulate tory comorbidities via a holistic approach, although this demand-adapted myelopoiesis in the bone marrow148. concept has to be addressed in future investigations. Furthermore, in the light of these emerging mech- anisms as bases of comorbidities (for example, trained Published online 28 January 2021

1. Genco, R. J. & Sanz, M. Clinical and public health causation and treatment with small-molecule markers, glycaemic control and vascular and kidney implications of periodontal and systemic diseases: inhibitors. Sci. Adv. 5, eaau3333 (2019). functions in patients with type 2 diabetes mellitus. an overview. Periodontol. 2000 83, 7–13 (2020). This study provides strong clinical and 20. Ishai, A. et al. Periodontal disease associates 2. Potempa, J., Mydel, P. & Koziel, J. The case for experimental evidence that the oral pathogen with arterial inflammation via potentiation of periodontitis in the pathogenesis of rheumatoid P. gingivalis can infect the brain and contribute a hematopoietic–arterial axis. JACC Cardiovasc. arthritis. Nat. Rev. Rheumatol. 13, 606–620 to inflammatory pathology in Alzheimer disease. Imaging 12, 2271–2273 (2019). (2017). 12. Konig, M. F. et al. Aggregatibacter An important clinical–imaging study that correlates 3. Hajishengallis, G. Periodontitis: from microbial actinomycetemcomitans-induced hypercitrullination periodontal metabolic activity/inflammation with immune subversion to systemic inflammation. links periodontal infection to autoimmunity in haematopoietic tissue activity and arterial Nat. Rev. Immunol. 15, 30–44 (2015). rheumatoid arthritis. Sci. Transl. Med. 8, 369ra176 inflammation, thus supporting the hypothesis 4. Acharya, C., Sahingur, S. E. & Bajaj, J. S. Microbiota, (2016). derived from experimental studies that the cirrhosis, and the emerging oral-gut-liver axis. In this study, the periodontal pathogen inflammatory adaptation of haematopoietic JCI Insight 2, e94416 (2017). A. actinomycetemcomitans is shown to induce the progenitor cells in the bone marrow may link 5. Schenkein, H. A., Papapanou, P. N., Genco, R. & generation of citrullinated autoantigens involved different chronic inflammatory diseases. Sanz, M. Mechanisms underlying the association in rheumatoid arthritis through the action of its 21. Teles, R. & Wang, C.-Y. Mechanisms involved in between periodontitis and atherosclerotic disease. pore-forming leukotoxin A (LtxA) on neutrophils. the association between peridontal diseases and Periodontol. 2000 83, 90–106 (2020). Anti-LtxA antibodies are strongly associated cardiovascular disease. Oral Dis. 17, 450–461 6. Kitamoto, S. et al. The intermucosal connection with ACPA positivity in patients with rheumatoid (2011). between the mouth and gut in commensal pathobiont- arthritis. 22. Bajaj, J. S. et al. Periodontal therapy favorably driven colitis. Cell 182, 447–462 (2020). 13. Meghil, M. M. et al. Disruption of immune homeostasis modulates the oral-gut-hepatic axis in cirrhosis. This study elegantly shows that oral bacterial- in human dendritic cells via regulation of autophagy Am. J. Physiol. Gastrointest. Liver Physiol. 315,

specific TH17 cells, which expand during and apoptosis by Porphyromonas gingivalis. G824–G837 (2018). experimental periodontitis, are instructed to migrate Front. Immunol. 10, 2286 (2019). 23. Xiao, E. et al. Diabetes enhances IL-17 expression and to the gut, where they are activated by translocated 14. Sato, K. et al. Aggravation of collagen-induced alters the oral microbiome to increase its pathogenicity. (via the gastrointestinal route) oral bacteria and arthritis by orally administered Porphyromonas Cell Host Microbe 22, 120–128 e124 (2017). contribute to the development of colitis. gingivalis through modulation of the gut microbiota 24. Teles, F., Wang, Y., Hajishengallis, G., Hasturk, H. & 7. Abed, J. et al. Fap2 mediates Fusobacterium and gut immune system. Sci. Rep. 7, 6955 (2017). Marchesan, J. Impact of systemic factors in shaping nucleatum colorectal adenocarcinoma enrichment 15. Farrugia, C. et al. Mechanisms of vascular damage the periodontal microbiome. Periodontol. 2000 85, by binding to tumor-expressed Gal-GalNAc. by systemic dissemination of the oral pathogen 126–160 (2021). Cell Host Microbe 20, 215–225 (2016). Porphyromonas gingivalis. FEBS J. https://doi.org/ 25. Monsarrat, P. et al. Clinical research activity in This paper explains F. nucleatum’s tropism to 10.1111/febs.15486 (2020). periodontal medicine: a systematic mapping of CRC. The bacterium expresses a Gal-GalNAc- 16. Zhao, Y. et al. Characterization and regulation of trial registers. J. Clin. Periodontol. 43, 390–400 binding lectin (Fap2) that can bind to the host osteoclast precursors following chronic Porphyromonas (2016). polysaccharide Gal-GalNAc, which is highly gingivalis infection. J. Leukoc. Biol. 108, 1037–1050 26. Rydén, L. et al. Periodontitis increases the risk of a expressed in human CRC. (2020). first myocardial infarction. Circulation 133, 576–583 8. Atarashi, K. et al. Ectopic colonization of oral This investigation identifies a mechanism whereby (2016). bacteria in the intestine drives TH1 cell induction the bone marrow may link periodontitis to other 27. Rodríguez-Lozano, B. et al. Association between and inflammation. Science 358, 359–365 (2017). bone loss disorders. Specifically, periodontal severity of periodontitis and clinical activity in This is the first paper to show that ectopic gut bacteria-induced serum IL-6 acts in the bone marrow rheumatoid arthritis patients: a case–control study. colonization by bacteria of oral origin results in to promote the expansion and osteoclastogenic Arthritis Res. Ther. 21, 27 (2019). expansion of colitogenic T cells and promotion lineage bias of a precursor population, which can 28. Genco, R. J. & Van Dyke, T. E. Prevention: reducing of colitis in a susceptible host. traffic to sites of bone resorption and differentiate the risk of CVD in patients with periodontitis. 9. Blasco-Baque, V. et al. Periodontitis induced by into mature osteoclasts. Nat. Rev. Cardiol. 7, 479–480 (2010). Porphyromonas gingivalis drives periodontal 17. Netea, M. G. et al. Defining trained immunity and 29. D’Aiuto, F., Orlandi, M. & Gunsolley, J. C. Evidence microbiota dysbiosis and insulin resistance via its role in health and disease. Nat. Rev. Immunol. 20, that periodontal treatment improves biomarkers and an impaired adaptive immune response. Gut 66, 375–388 (2020). CVD outcomes. J. Clin. Periodontol. 40, S85–S105 872–885 (2017). 18. Chavakis, T., Mitroulis, I. & Hajishengallis, G. (2013). 10. Gur, C. et al. Binding of the Fap2 protein of Hematopoietic progenitor cells as integrative hubs 30. H. Bokhari, S. A. et al. Non-surgical periodontal to human inhibitory for adaptation to and fine-tuning of inflammation. therapy reduces coronary heart disease risk markers: receptor TIGIT protects tumors from immune cell Nat. Immunol. 20, 802–811 (2019). a randomized controlled trial. J. Clin. Periodontol. 39, attack. Immunity 42, 344–355 (2015). 19. D’Aiuto, F. et al. Systemic effects of periodontitis 1065–1074 (2012). This study identifies an important immune evasion treatment in patients with type 2 diabetes: 31. de Oliveira, C., Watt, R. & Hamer, M. Toothbrushing, mechanism by F. nucleatum, which uses its Fap2 a 12 month, single-centre, investigator-masked, inflammation, and risk of cardiovascular disease: adhesin to activate the inhibitory immunoreceptor randomised trial. Lancet Diabetes Endocrinol. 6, results from Scottish health survey. Br. Med. J. 340, TIGIT, which disables the tumour-killing activity of 954–965 (2018). c2451 (2010). T cells and natural killer cells. A well-designed and executed randomized 32. Tonetti, M. S. Periodontitis and risk for atherosclerosis: 11. Dominy, S. S. et al. Porphyromonas gingivalis in controlled trial showing favourable effects of local an update on intervention trials. J. Clin. Periodontol. Alzheimer’s disease brains: evidence for disease periodontal treatment on systemic inflammatory 36, 15–19 (2009).

438 | July 2021 | volume 21 www.nature.com/nri Reviews

33. Türer Ç, C., Durmuş, D., Balli, U. & Güven, B. ligature-induced periodontitis in rats. J. Periodontol. associates with a type 1 IFN signature in peripheral Effect of non-surgical periodontal treatment on 88, e49–e57 (2017). blood neutrophils. J. Immunol. 181, 5775–5784 gingival crevicular fluid and serum endocan, vascular 60. Tsukasaki, M. et al. Host defense against oral (2008). endothelial growth factor-A, and tumor necrosis microbiota by bone-damaging T cells. Nat. Commun. 81. Kalafati, L. et al. Innate immune training of factor-alpha levels. J. Periodontol. 88, 493–501 9, 701 (2018). granulopoiesis promotes anti-tumor activity. Cell 183, (2017). 61. Komazaki, R. et al. Periodontal pathogenic bacteria, 771–785 (2020). 34. Mammen, M. J., Scannapieco, F. A. & Sethi, S. Aggregatibacter actinomycetemcomitans affect 82. Rafferty, B. et al. Impact of monocytic cells on recovery Oral–lung microbiome interactions in lung diseases. non-alcoholic fatty liver disease by altering gut of uncultivable bacteria from atherosclerotic lesions. Periodontol. 2000 83, 234–241 (2020). microbiota and glucose metabolism. Sci. Rep. 7, J. Intern. Med. 270, 273–280 (2011). 35. Qin, N. et al. Alterations of the human gut microbiome 13950 (2017). 83. Kozarov, E. V., Dorn, B. R., Shelburne, C. E., in liver cirrhosis. Nature 513, 59–64 (2014). 62. King, K. Y. & Goodell, M. A. Inflammatory modulation Dunn, W. A. Jr & Progulske-Fox, A. Human 36. Arimatsu, K. et al. Oral pathobiont induces systemic of HSCs: viewing the HSC as a foundation for the atherosclerotic plaque contains viable invasive inflammation and metabolic changes associated with immune response. Nat. Rev. Immunol. 11, 685 Actinobacillus actinomycetemcomitans and alteration of gut microbiota. Sci. Rep. 4, 4828 (2014). (2011). Porphyromonas gingivalis. Arterioscler. Thromb. 37. Hotamisligil, G. S. Inflammation and metabolic 63. Yamamoto, R. et al. Clonal analysis unveils self- Vasc. Biol. 25, e17–e18 (2005). disorders. Nature 444, 860–867 (2006). renewing lineage-restricted progenitors generated 84. Carrion, J. et al. Microbial carriage state of peripheral 38. Norata, G. D. et al. The cellular and molecular basis directly from hematopoietic stem cells. Cell 154, blood dendritic cells (DCs) in of translational immunometabolism. Immunity 43, 1112–1126 (2013). influences DC differentiation, atherogenic potential. 421–434 (2015). 64. Tsukasaki, M. & Takayanagi, H. Osteoimmunology: J. Immunol. 189, 3178–3187 (2012). 39. Orlandi, M., Graziani, F. & D’Aiuto, F. Periodontal evolving concepts in bone-immune interactions in 85. Gevers, D. et al. The treatment-naive microbiome in therapy and cardiovascular risk. Periodontol. 2000 health and disease. Nat. Rev. Immunol. 19, 626–642 new-onset Crohn’s disease. Cell Host Microbe 15, 83, 107–124 (2020). (2019). 382–392 (2014). 40. Genco, R. J., Graziani, F. & Hasturk, H. Effects of 65. Herrera, B. S. et al. Peripheral blood mononuclear 86. Schirmer, M. et al. Compositional and temporal periodontal disease on glycemic control, phagocytes from patients with chronic periodontitis changes in the gut microbiome of pediatric ulcerative complications, and incidence of diabetes mellitus. are primed for osteoclast formation. J. Periodontol. colitis patients are linked to disease course. Cell Host Periodontol. 2000 83, 59–65 (2020). 85, e72–e81 (2014). Microbe 24, 600–610.e604 (2018). 41. Jepsen, S., Suvan, J. & Deschner, J. The association 66. Manz, M. G. & Boettcher, S. Emergency Gevers et al. and Schirmer et al. demonstrate that of periodontal diseases with metabolic syndrome and granulopoiesis. Nat. Rev. Immunol. 14, 302–314 the dominant species with increased abundance in obesity. Periodontol. 2000 83, 125–153 (2020). (2014). paediatric patients with IBD are actually derived 42. Tonetti, M. S. et al. Treatment of periodontitis and 67. Arts, R. J. W. et al. BCG vaccination protects against from the oral cavity instead of being typical endothelial function. N. Engl. J. Med. 356, 911–920 experimental viral infection in humans through the intestinal bacteria. (2007). induction of cytokines associated with trained 87. Lamont, R. J., Koo, H. & Hajishengallis, G. The 43. Desvarieux, M. et al. Changes in clinical and immunity. Cell Host Microbe 23, 89–100 (2018). oral microbiota: dynamic communities and host microbiological periodontal profiles relate to 68. Mitroulis, I. et al. Modulation of myelopoiesis interactions. Nat. Rev. Microbiol. 16, 745–759 progression of carotid intima-media thickness: progenitors is an integral component of trained (2018). the oral infections and vascular disease epidemiology immunity. Cell 172, 147–161 (2018). 88. Gimbrone, M. A. Jr & García-Cardeña, G. study. J. Am. Heart Assoc. 2, e000254 (2013). This study shows that trained immunity can be Endothelial cell dysfunction and the pathobiology of 44. Yoneda, M. et al. Involvement of a periodontal initiated in the bone marrow through long-lasting atherosclerosis. Circ. Res. 118, 620–636 (2016). pathogen, Porphyromonas gingivalis on the metabolic and transcriptional adaptations in 89. Mougeot, J. C. et al. Porphyromonas gingivalis is pathogenesis of non-alcoholic fatty liver disease. HSPCs that lead to enhanced myelopoiesis. the most abundant species detected in coronary BMC Gastroenterol. 12, 16 (2012). 69. Pietras, E. M. et al. Chronic interleukin-1 exposure and femoral arteries. J. Oral Microbiol. 9, 1281562 45. Helenius-Hietala, J. et al. Periodontitis is associated drives haematopoietic stem cells towards precocious (2017). with incident chronic liver disease — a population- myeloid differentiation at the expense of self-renewal. 90. Dioguardi, M. et al. The role of periodontitis and based cohort study. Liver Int. 39, 583–591 (2019). Nat. Cell Biol. 18, 607–618 (2016). periodontal bacteria in the onset and progression of 46. Brito, L. C. W. et al. Experimental periodontitis 70. Fifer, K. M. et al. Positron emission tomography Alzheimer’s disease: a systematic review. J. Clin. Med. promotes transient vascular inflammation and measurement of periodontal 18F-fluorodeoxyglucose 9, 495 (2020). endothelial dysfunction. Arch. Oral. Biol. 58, uptake is associated with histologically determined 91. Beydoun, M. A. et al. Clinical and bacterial markers 1187–1198 (2013). carotid plaque inflammation. J. Am. Coll. Cardiol. 57, of periodontitis and their association with incident 47. Matsuda, Y. et al. Ligature-induced periodontitis 971–976 (2011). all-cause and Alzheimer’s disease dementia in a large in mice induces elevated levels of circulating 71. Ling, M. R., Chapple, I. L. & Matthews, J. B. Peripheral national survey. J. Alzheimers Dis. 75, 157–172 interleukin-6 but shows only weak effects on adipose blood neutrophil cytokine hyper-reactivity in chronic (2020). and liver tissues. J. Periodont. Res. 51, 639–646 periodontitis. Innate Immun. 21, 714–725 (2015). 92. Kovacech, B. & Novak, M. Tau truncation is a (2016). 72. Radvar, M., Tavakkol-Afshari, J., Bajestan, M. N., productive posttranslational modification of 48. O’Boyle, C. et al. Ligature-induced periodontitis Naseh, M. R. & Arab, H. R. The effect of periodontal neurofibrillary degeneration in Alzheimer’s disease. induces systemic inflammation but does not alter treatment on IL-6 production of peripheral blood Curr. Alzheimer Res. 7, 708–716 (2010). acute outcome after stroke in mice. Int. J. Stroke 15, monocytes in and chronic 93. Long, J. M. & Holtzman, D. M. Alzheimer disease: 175–187 (2019). periodontitis patients. Iran. J. Immunol. 5, 100–106 an update on pathobiology and treatment strategies. 49. Anbinder, A. L. et al. Periodontal disease exacerbates (2008). Cell 179, 312–339 (2019). systemic ovariectomy-induced bone loss in mice. Bone 73. Kleinnijenhuis, J. et al. Bacille Calmette–Guerin 94. Muñoz, S. S., Garner, B. & Ooi, L. Understanding 83, 241–247 (2016). induces NOD2-dependent nonspecific protection the role of ApoE fragments in Alzheimer’s disease. 50. Miyajima, S.-i. et al. Periodontitis-activated from reinfection via epigenetic reprogramming Neurochem. Res. 44, 1297–1305 (2019). monocytes/macrophages cause aortic inflammation. of monocytes. Proc. Natl Acad. Sci. USA 109, 95. Lönn, J. et al. Lipoprotein modifications by gingipains Sci. Rep. 4, 5171 (2014). 17537–17542 (2012). of Porphyromonas gingivalis. J. Periodont. Res. 53, 51. Hasturk, H. et al. Resolvin E1 (RvE1) attenuates 74. Cirovic, B. et al. BCG vaccination in humans elicits 403–413 (2018). atherosclerotic plaque formation in diet and trained immunity via the hematopoietic progenitor 96. Poole, S. et al. Active invasion of Porphyromonas inflammation-induced atherogenesis. Arterioscl. compartment. Cell Host Microbe 28, 322–334 gingivalis and infection-induced complement Thromb. Vasc. Biol. 35, 1123–1133 (2015). (2020). activation in ApoE–/– mice brains. J. Alzheimers Dis. 52. Tian, J. et al. Porphyromonas gingivalis induces This study shows that BCG vaccination in humans 43, 67–80 (2015). insulin resistance by increasing BCAA levels in mice. induces sustained transcriptomic myeloid bias in 97. Hajishengallis, G., Reis, E. S., Mastellos, D. C., J. Dent. Res. 99, 839–846 (2020). HSPCs associated with long-term heightened Ricklin, D. & Lambris, J. D. Novel mechanisms 53. White, P. J. & Newgard, C. B. Branched-chain amino responsiveness of circulating myeloid cells to and functions of complement. Nat. Immunol. 18, acids in disease. Science 363, 582–583 (2019). inflammatory stimuli, thereby providing human 1288–1298 (2017). 54. Chavakis, T. et al. The pattern recognition receptor relevance for the findings of Mitroulis et al. (2018). 98. Yin, C. et al. ApoE attenuates unresolvable (RAGE) is a counterreceptor for leukocyte integrins: 75. Bernelot Moens, S. J. et al. Unexpected arterial inflammation by complex formation with activated a novel pathway for inflammatory cell recruitment. wall and cellular inflammation in patients with C1q. Nat. Med. 25, 496–506 (2019). J. Exp. Med. 198, 1507–1515 (2003). rheumatoid arthritis in remission using biological 99. Kantarci, A. et al. Combined administration of resolvin 55. Ruiz, H. H., Ramasamy, R. & Schmidt, A. M. Advanced therapy: a cross-sectional study. Arthritis Res. Ther. E1 and lipoxin A4 resolves inflammation in a murine glycation end products: building on the concept of the 18, 115 (2016). model of Alzheimer’s disease. Exp. Neurol. 300, “Common Soil” in metabolic disease. Endocrinology 76. Schloss, M. J., Swirski, F. K. & Nahrendorf, M. 111–120 (2018). 161, 1–10 (2020). Modifiable cardiovascular risk, hematopoiesis, and 100. Kantarci, A. et al. Microglial response to experimental 56. Lalla, E. et al. Blockade of RAGE suppresses innate immunity. Circ. Res. 126, 1242–1259 (2020). periodontitis in a murine model of Alzheimer’s periodontitis-associated bone loss in diabetic mice. 77. Christ, A. et al. Western diet triggers NLRP3- disease. Sci. Rep. 10, 18561 (2020). J. Clin. Invest. 105, 1117–1124 (2000). dependent innate immune reprogramming. Cell 172, 101. Papageorgiou, S. N. et al. Inflammatory bowel disease 57. Nakahara, T. et al. Involvement of Porphyromonas 162–175 (2018). and oral health: systematic review and a meta-analysis. gingivalis in the progression of non-alcoholic fatty 78. Bekkering, S. et al. Trained immunity: linking obesity J. Clin. Periodontol. 44, 382–393 (2017). liver disease. J. Gastroenterol. 53, 269–280 (2018). and cardiovascular disease across the life-course? 102. Pietropaoli, D. et al. Occurrence of spontaneous 58. Nagasaki, A. et al. Odontogenic infection by Trends Endocrinol. Metab. 31, 378–389 (2020). periodontal disease in the SAMP1/YitFc Murine model Porphyromonas gingivalis exacerbates fibrosis in 79. Pink, C. et al. Longitudinal effects of systemic of Crohn disease. J. Periodontol. 85, 1799–1805 NASH via hepatic stellate cell activation. Sci. Rep. 10, inflammation markers on periodontitis. J. Clin. (2014). 4134 (2020). Periodontol. 42, 988–997 (2015). 103. Moran, C. J. et al. IL-10R polymorphisms are 59. Vasconcelos, D. F. P. et al. Decrease of pericytes is 80. Wright, H. J., Matthews, J. B., Chapple, I. L., associated with very-early-onset ulcerative colitis. associated with liver disease caused by Ling-Mountford, N. & Cooper, P. R. Periodontitis Inflamm. Bowel Dis. 19, 115–123 (2013).

NATuRe RevIewS | IMMunOlOgy volume 21 | July 2021 | 439 Reviews

104. Schmidt, T. S. et al. Extensive transmission of peptidylarginine deiminase (PAD). PLoS Pathog. 9, 144. Eskan, M. A. et al. The leukocyte integrin antagonist microbes along the gastrointestinal tract. eLife 8, e1003627 (2013). Del-1 inhibits IL-17-mediated inflammatory bone loss. e42693 (2019). 126. Gully, N. et al. Porphyromonas gingivalis Nat. Immunol. 13, 465–473 (2012). 105. Yachida, S. et al. Metagenomic and metabolomic peptidylarginine deiminase, a key contributor in the 145. Shin, J. et al. DEL-1 restrains osteoclastogenesis analyses reveal distinct stage-specific phenotypes of pathogenesis of experimental periodontal disease and and inhibits inflammatory bone loss in nonhuman the gut microbiota in colorectal cancer. Nat. Med. 25, experimental arthritis. PLoS ONE 9, e100838 (2014). primates. Sci. Transl. Med. 7, 307ra155 (2015). 968–976 (2019). Maresz et al. is the first study to mechanistically 146. Kourtzelis, I. et al. DEL-1 promotes macrophage 106. Thomas, A. M. et al. Metagenomic analysis of link a periodontal pathogen (P. gingivalis) with efferocytosis and clearance of inflammation. colorectal cancer datasets identifies cross-cohort generation of citrullinated autoantigens that Nat. Immunol. 20, 40–49 (2019). microbial diagnostic signatures and a link with choline are involved in the pathogenesis of rheumatoid 147. Li, X. et al. The DEL-1/β3 integrin axis promotes degradation. Nat. Med. 25, 667–678 (2019). arthritis. Gully et al. provide independent regulatory T cell responses during inflammation 107. Kostic, A. D. et al. Fusobacterium nucleatum confirmation that peptidylarginine deiminase of resolution. J. Clin. Invest. 130, 6261–6277 potentiates intestinal tumorigenesis and modulates P. gingivalis promotes the induction of ACPAs and (2020). the tumor-immune microenvironment. Cell Host experimental arthritis. 148. Mitroulis, I. et al. Secreted protein Del-1 regulates Microbe 14, 207–215 (2013). 127. Wegner, N. et al. Peptidylarginine deiminase from myelopoiesis in the hematopoietic stem cell niche. 108. Wirbel, J. et al. Meta-analysis of fecal metagenomes Porphyromonas gingivalis citrullinates human J. Clin. Invest. 127, 3624–3639 (2017). reveals global microbial signatures that are specific fibrinogen and alpha-enolase: implications for 149. Adler, C. J. et al. Sequencing ancient calcified dental for colorectal cancer. Nat. Med. 25, 679–689 (2019). autoimmunity in rheumatoid arthritis. Arthritis plaque shows changes in oral microbiota with dietary 109. Castellarin, M. et al. Fusobacterium nucleatum Rheumatol. 62, 2662–2672 (2010). shifts of the Neolithic and Industrial revolutions. infection is prevalent in human colorectal carcinoma. 128. Lübcke, P. M. et al. Periodontal treatment prevents Nat. Genet. 45, 450–455 (2013). Genome Res. 22, 299–306 (2012). arthritis in mice and methotrexate ameliorates 150. Peres, M. A. et al. Oral diseases: a global public health 110. Kostic, A. D. et al. Genomic analysis identifies periodontal bone loss. Sci. Rep. 9, 8128 (2019). challenge. Lancet 394, 249–260 (2019). association of Fusobacterium with colorectal 129. Flak, M. B. et al. Inflammatory arthritis disrupts gut 151. Dutzan, N. et al. A dysbiotic microbiome triggers carcinoma. Genome Res. 22, 292–298 (2012). resolution mechanisms, promoting barrier breakdown TH17 cells to mediate oral mucosal immunopathology 111. Figuero, E., Han, Y. W. & Furuichi, Y. Periodontal by Porphyromonas gingivalis. JCI Insight 4, e125191 in mice and humans. Sci. Transl. Med. 10, eaat0797 diseases and adverse pregnancy outcomes: (2019). (2018). mechanisms. Periodontol. 2000 83, 175–188 (2020). 130. Boyaka, P. N. & Fujihashi, K. in Clinical Immunology 152. Abe, T. & Hajishengallis, G. Optimization of the 112. Rubinstein, M. R. et al. Fusobacterium nucleatum (Fifth Edition) (eds Rich, R. R. et al.) 285–298 ligature-induced periodontitis model in mice. promotes colorectal carcinogenesis by modulating (Elsevier, 2019). J. Immunol. Methods 394, 49–54 (2013). E-cadherin/β-catenin signaling via its FadA adhesin. 131. Gill, N., Wlodarska, M. & Finlay, B. B. The future 153. Hajishengallis, G. et al. Low-abundance biofilm Cell Host Microbe 14, 195–206 (2013). of mucosal immunology: studying an integrated species orchestrates inflammatory periodontal 113. Casasanta, M. A. et al. Fusobacterium nucleatum system-wide organ. Nat. Immunol. 11, 558–560 disease through the commensal microbiota and host-cell binding and invasion induces IL-8 and CXCL1 (2010). complement. Cell Host Microbe 10, 497–506 secretion that drives colorectal cancer cell migration. 132. Scribano, M. L. Vedolizumab for inflammatory bowel (2011). Sci. Signal. 13, eaba9157 (2020). disease: from randomized controlled trials to real-life 154. Payne, M. A. et al. Horizontal and vertical transfer This study identifies a new mechanism linking evidence. World J. Gastroenterol. 24, 2457–2467 of oral microbial dysbiosis and periodontal disease. F. nucleatum to CRC by showing that this oral (2018). J. Dent. Res. 98, 1503–1510 (2019). bacterium promotes chemokine secretion that 133. Calderón-Gómez, E. et al. Commensal-specific CD4+ 155. Penkov, S., Mitroulis, I., Hajishengallis, G. & stimulates tumour cell migration and invasion. cells from patients with Crohn’s disease have a Chavakis, T. Immunometabolic crosstalk: an ancestral 114. Yu, T. et al. Fusobacterium nucleatum promotes T-helper 17 inflammatory profile. Gastroenterology principle of trained immunity? Trends Immunol. 40, chemoresistance to colorectal cancer by modulating 151, 489–500 (2016). 1–11 (2019). autophagy. Cell 170, 548–563 (2017). 134. Wang, J. et al. Respiratory influenza virus infection 156. Kaufmann, E. et al. BCG educates hematopoietic stem This paper shows that F. nucleatum adversely induces intestinal immune injury via microbiota- cells to generate protective innate immunity against affects the CRC chemotherapeutic response by mediated Th17 cell–dependent inflammation. tuberculosis. Cell 172, 176–190 (2018). modulating autophagy. J. Exp. Med. 211, 2397–2410 (2014). 157. Nahrendorf, M. Myeloid cell contributions to 115. Komiya, Y. et al. Patients with colorectal cancer have 135. D’Amico, F., Baumgart, D. C., Danese, S. & cardiovascular health and disease. Nat. Med. 24, identical strains of Fusobacterium nucleatum in their Peyrin-Biroulet, L. Diarrhea during COVID-19 711–720 (2018). colorectal cancer and oral cavity. Gut 68, 1335–1337 infection: pathogenesis, epidemiology, prevention, 158. Murphy, A. J. & Tall, A. R. Disordered haematopoiesis (2019). and management. Clin. Gastroenterol. Hepatol. 18, and athero-thrombosis. Eur. Heart J. 37, 1113–1121 116. Abed, J. et al. Colon cancer-associated Fusobacterium 1663–1672 (2020). (2016). nucleatum may originate from the oral cavity and 136. Miles, B. et al. Secondary lymphoid organ homing 159. Barrett, T. J., Murphy, A. J., Goldberg, I. J. & reach colon tumors via the circulatory system. phenotype of human myeloid dendritic cells disrupted Fisher, E. A. Diabetes-mediated myelopoiesis and Front. Cell Infect. Microbiol. 10, 400 (2020). by an intracellular oral pathogen. Infect. Immun. 82, the relationship to cardiovascular risk. Ann. N. Y. 117. Ning, Y. & Lenz, H. J. Targeting IL-8 in colorectal 101–111 (2014). Acad. Sci. 1402, 31–42 (2017). cancer. Expert. Opin. Ther. Targets 16, 491–497 137. Sainz, J. & Sata, M. CXCR4, a key modulator of 160. Tall, A. R. & Yvan-Charvet, L. Cholesterol, inflammation (2012). vascular progenitor cells. Arterioscler. Thromb. and innate immunity. Nat. Rev. Immunol. 15, 118. Rubinstein, M. R. et al. Fusobacterium nucleatum Vasc. Biol. 27, 263–265 (2007). 104–116 (2015). promotes colorectal cancer by inducing Wnt/β-catenin 138. Rajendran, M. et al. Systemic antibiotic therapy modulator annexin A1. EMBO Rep 20, e47638 reduces circulating inflammatory dendritic cells and Acknowledgements (2019). Treg-Th17 plasticity in periodontitis. J. Immunol. 202, The authors’ research is supported by grants from the US 119. Sun, J. et al. Role of the oral microbiota in 2690–2699 (2019). National Institutes of Health (DE024153, DE024716, cancer evolution and progression. Cancer Med. 9, 139. Chen, M. et al. Dendritic cell apoptosis in the DE029436 to G.H.; DE026152 and DE028561 to G.H. and 6306–6321 (2020). maintenance of immune tolerance. Science 311, T.C.) and the German Research Foundation (CRC-TR127 120. Hashimoto, M. et al. Periodontitis and Porphyromonas 1160–1164 (2006). and CRC1181 to T.C.). The authors regret that several gingivalis in preclinical stage of arthritis patients. 140. Hajishengallis, G., Chavakis, T. & Lambris, J. D. important studies could only be cited indirectly through com- PLoS ONE 10, e0122121 (2015). Current understanding of periodontal disease prehensive reviews, owing to space and reference number 121. Mikuls, T. R. et al. Periodontitis and Porphyromonas pathogenesis and targets for host-modulation therapy. limitations. gingivalis in patients with rheumatoid arthritis. Periodontol. 2000 84, 14–34 (2020). Arthritis Rheumatol. 66, 1090–1100 (2014). 141. Ridker, P. M. et al. Antiinflammatory therapy with Author contributions 122. Al-Katma, M. K., Bissada, N. F., Bordeaux, J. M., canakinumab for atherosclerotic disease. N. Engl. The authors contributed equally to all aspects of the article. Sue, J. & Askari, A. D. Control of periodontal infection J. Med. 377, 1119–1131 (2017). reduces the severity of active rheumatoid arthritis. 142. Jaiswal, S. & Libby, P. Clonal haematopoiesis: Competing interests J. Clin. Rheumatol. 13, 134–137 (2007). connecting ageing and inflammation in cardiovascular The authors declare no competing interests. 123. Khare, N. et al. Nonsurgical periodontal therapy disease. Nat. Rev. Cardiol. 17, 137–144 (2020). decreases the severity of rheumatoid arthritis: 143. Fuster, J. J. et al. Clonal hematopoiesis associated Peer review information a case-control study. J. Contemp. Dent. Pract. 17, with TET2 deficiency accelerates atherosclerosis Nature Reviews Immunology thanks N. Kamada, T. Van Dyke 484–488 (2016). development in mice. Science 355, 842–847 (2017). and the other, anonymous, reviewer(s) for their contribution 124. Ortiz, P. et al. Periodontal therapy reduces the This study provides causal evidence linking to the peer review of this work. severity of active rheumatoid arthritis in patients CHIP to atherosclerosis. The authors show that treated with or without tumor necrosis factor atherosclerosis progresses more rapidly in mice Publisher’s note inhibitors. J. Periodontol. 80, 535–540 (2009). transplanted with TET2-deficient bone marrow Springer Nature remains neutral with regard to jurisdictional 125. Maresz, K. J. et al. Porphyromonas gingivalis cells, in great part attributed to the increased claims in published maps and institutional affiliations. facilitates the development and progression of inflammatory activity of TET2-deficient destructive arthritis through its unique bacterial macrophage progeny. © Springer Nature Limited 2021

440 | July 2021 | volume 21 www.nature.com/nri